Steady-state groundwater flow and contaminant

transport modelling
of Akaki wellfield and its surrounding
catchment
(Addis Ababa, Ethiopia)

Alema Tesfaye

March, 2009
Steady-state groundwater flow and contaminant transport
modelling
of Akaki wellfield and its surrounding catchment
(Addis Ababa, Ethiopia)
by

Alema Tesfaye

Thesis submitted to the International Institute for Geo-information Science and Earth Observation in
partial fulfilment of the requirements for the degree of Master of Science in Geo-information Science
and Earth Observation, Specialisation: (Groundwater Assessment and Modelling)

Thesis Assessment Board

Chairman Dr.Ir.M.W.Lubczynski WRS, ITC, Enschede

External Examiner Dr.Ir.P.Droogers Future Water, Wagingen
First Supervisor Dr.A .S .M. Gieske WRS, ITC, Enschede
Second Supervisor Dr. T.H.M. Tom Rientjes WRS, ITC, Enschede

INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE AND EARTH OBSERVATION

ENSCHEDE, THE NETHERLANDS
 Disclaimer

This document describes work undertaken as part of a programme of study at the International
Institute for Geo-information Science and Earth Observation. All views and opinions expressed
therein remain the sole responsibility of the author, and do not necessarily represent those of
the institute.
 Dedicated to
My Parents with Love and Gratitude
“Who encourage me to knowledge”
Abstract
The study focuses on groundwater modelling of the Akaki catchment and the Akaki wellfield in
particular under steady state flow conditions. Besides that, advective contaminant transport modelling
was implemented in Akaki wellfield to delineate capture zone. The aquifer system was modelled
using PMWIN (Chiang and Kinzelbach, 1998) as pre and post processor for MODFLOW (McDonald
and Harbaugh, 1988). The aquifer was simulated under confined/unconfined condition and is
represented by a single layer of 100m constant thickness. The grid cells of the model were taken
400x400m for Akaki catchment with the area of 1500 km2. A grid size of 400x200m was used to
represent the wellfield which is approximated 197.3 km2 in extent. Model input parameters were
determined based on analysis of pumping test data.

The recharge condition in Akaki catchment is characterized by three sources of water: rain recharge,
the river bed infiltration and the seepage from Aba Samuel and Legedadi lakes. In addition, the
recharge conditions in Akaki wellfield include a fourth component which is inflow from the northern
and north-western boundary. Outflow (discharge) conditions from the aquifer are characterized as
base flow, springs, subsurface lateral flow and well abstractions. Boundary conditions were assigned
to the model domain to simulate inflow and outflow terms of the model domain. The Chloride Mass
Balance Method (CMB), semi-distributed water balance model and recursive digital filter methods
were employed to estimate the recharge in the catchment. From the water balance as simulated by the
catchment’s model, the total average base flow estimated is 92 MCM/yr or 62 mm/yr. From the water
balance model described in chapter 3, it is found that a base flow and recharge value of 130 MCM/yr
or 87 mm/yr is found. Using the digital filter method, the base flow and recharge values are roughly
65 MCM/yr (63mm/yr) indicating good agreement between the models.
The trial and error method was used to calibrate the models using the observed and simulated
hydraulic heads. The water budget of the area reached equilibrium conditions with recharge from
precipitation 158 MCM/yr, well abstraction some 23 MCM/yr, 8 MCM/yr as out flow from the
catchment at the catchment outlet, 130 MCM/yr seepage from aquifer to the surface water bodies and
11 MCM/yr from surface water bodies to the aquifer. Optimised parameters (hydraulic conductivity
and recharge) are spatially distributed over the model area.

Furthermore, PMPATH, has been used at the wellfield eventually to calculate path lines and travel
times of contamination. This approach involved the introduction of particles at contaminant sources
upstream of wells and at the wellfield itself, then identifying the path lines, and finally determining
the spatial distribution of the contaminants through steady state flow field. The flow lines converge
towards Akaki wellfield from all directions, implying that any contaminated water from the upper part
of the aquifer will end in the wells, indicating a high risk of vulnerability of the wellfield to pollution.
The following recommendations are helpful in curing the risks posed. Manufacturing activities having
pollution potential must be limited in special areas sufficiently far from water supply wells; the
chemical quality of groundwater must be monitored and an environmental policy must be
implemented with particular emphasis on the protection zones around the wellfield.
Key Words: Addis Ababa - Akaki - Modelling - Modflow - Volcanic aquifer - Groundwater -
PMPATH

i
Acknowledgements
I would like to express my Sincere and heartfelt gratitude to the Netherlands Government through the
Netherlands Fellowship Program (NFP) for granting me the opportunity to pursue this course of study
without which I would not have realized my dream to further my studies. I am grateful to my
employer, Geological Survey of Ethiopia, for supporting me to fulfil my wish.

My thanks go to all the staff of WREM for the support and guidance throughout the modules and
thesis preparation.
I am greatly indebted to my first supervisor Dr. Ambro S.M. Gieske for his excellent guidance and
encouragements throughout my study period and especially during the research work. I highly
appreciate his constructive criticism and valuable advises which helped me to locate this research in
the right direction. I benefited a lot from discussions I had with him owing to which I gained a deeper
insight into and understanding of the factors governing groundwater occurrence and movement.
My gratitude also goes to my second supervisor Dr. T.H.M. Tom Rientjes, for laying me a sound
foundation on modelling during my study, kind guidance, valuable comments from the start of title
selection and field preparation up to end of my study.

I would like to also to express my appreciation to Ir. Arno Van Lieshout, the program director, for his
support and care.

My special thanks extend to Dr. Ir.M.W.Lubczynski for teaching me constructive and core ideas of
groundwater which was so useful for my research accomplishment. I acknowledge the support during
my laboratory work to Boudewijn de Smeth and Remco Dost.

I also wish to express my appreciation to all my class mates for their friendship, support, socialization
and help each other in times of pressure and stress. They were my new family during my stay and it
was a pleasure to be a member of them.

The Addis Ababa University, Addis Ababa Water and Sewerage Authority, Ethiopian Meteorological
Agency and Ministry of Water resources are greatly appreciated for the cooperation they showed in
providing data and information. I am indebted to my profession colleagues and friends particularily
Leta Gudisa, Frew, Feven Solomon, Mr. Shumet Kebede, Mr. Sheferaw Lulu, Fanuel Seyoum, and
Yonas Mulugeta who helped me a lot in the secondary data collection during my fieldwork. I
particularly appreciate Frew who dedicated his time and gave me a lot of helpful information during
my stay in the field work.

To my unique entire family and relatives no words can express my deep feeling of appreciation for
their valuable support through out my academic life for always being there for me.

ii
Table of contents
1. Introduction ......................................................................................................................... 1
1.1. General..................................................................................................................................1
1.2. Statement of the problem .....................................................................................................2
1.3. Objectives .............................................................................................................................2
1.3.1. General objective ................................................................................................................... 2
1.3.2. Specific objectives ................................................................................................................. 2
1.4. Research questions ...............................................................................................................3
1.5. Research hypothesis .............................................................................................................3
1.6. Previous works......................................................................................................................3
1.7. Methodology.........................................................................................................................4
1.7.1. Pre fieldwork.......................................................................................................................... 4
1.7.2. Field work .............................................................................................................................. 4
1.7.3. Post field work (data processing and analysis)....................................................................... 5
1.7.4. Methods to apply PMPATH (Advective transport model)..................................................... 5
1.8. Thesis outline........................................................................................................................8
2. The study area...................................................................................................................... 9
2.1. Location, digital elevation model and climate......................................................................9
2.1.1. Digital elevation model ........................................................................................................ 11
2.1.2. Climate ................................................................................................................................. 11
2.1.3. Precipitation ......................................................................................................................... 11
2.1.4. Temperature......................................................................................................................... 14
2.2. Hydrology ...........................................................................................................................15
2.3. Geological and hydrogeological setting .............................................................................16
2.3.1. Geology ............................................................................................................................... 17
2.3.2. Hydrogeology ...................................................................................................................... 17
2.4. Geomorphology and drainage system................................................................................19
2.5. Hydrochemistry ..................................................................................................................20
2.5.1. Water quality parameters ..................................................................................................... 21
2.5.2. Field sampling procedure ..................................................................................................... 21
2.5.3. Chemical analysis results...................................................................................................... 22
2.5.4. Reliability check................................................................................................................... 22
2.5.5. Presentation of results.......................................................................................................... 24
2.5.6. Water type deduction........................................................................................................... 24
2.6. Subsurface conceptualization of the model domains .........................................................26
2.6.1. Well log correlations ............................................................................................................ 26
2.6.2. Well test analysis.................................................................................................................. 32
2.6.3. Water level and abstraction monitoring................................................................................ 32
3. Groundwater flow modelling of Akaki catchment............................................................ 37
3.1. Introduction ........................................................................................................................37
3.2. Development of conceptual model .....................................................................................37
3.3. General Assumptions..........................................................................................................38
3.4. Governing Flow Equations .................................................................................................38
3.5. Modelling Approach...........................................................................................................39
3.6. Geographic Extension of the model ..................................................................................39

iii
 3.7. Spatial discretization of Model Grid ................................................................................. 39
3.8. Model structure.................................................................................................................. 40
3.9. Physical parameters of the system ..................................................................................... 41
3.9.1. Initial hydraulic heads .......................................................................................................... 41
3.9.2. Hydraulic conductivity ......................................................................................................... 42
3.9.2.1. Horizontal Hydraulic Conductivity....................................................................... 42
3.9.2.2. Vertical Hydraulic Conductivity or Leakance....................................................... 43
3.9.3. Horizontal Anisotropy.......................................................................................................... 43
3.9.4. Aquifer geometry.................................................................................................................. 43
3.9.5. Aquifer thickness.................................................................................................................. 44
3.9.6. Transmissivity ...................................................................................................................... 45
3.9.7. Effective Porosity................................................................................................................. 47
3.9.8. Boreholes and observation wells .......................................................................................... 47
3.9.9. Model Boundary Conditions ................................................................................................ 47
3.9.10. Sinks and sources of the modelled area................................................................................ 51
3.9.11. Recharge............................................................................................................................... 51
3.9.11.1. Groundwater recharge estimation by chloride mass balance method (CMB) ..... 52
3.9.11.2. Basic assumptions of chloride mass balance....................................................... 53
3.9.11.3. Baseflow separation: digital filter methods (recharge from baseflow)................ 55
3.9.11.4. Recharge estimation using semi-distributed water balance model ...................... 56
4. Calibration and execution of Akaki catchment model ......................................................63
4.1. Calibration of the model .................................................................................................... 63
4.2. Calibration target and uncertainty ..................................................................................... 63
4.2.1. Steady State Flow Calibration.............................................................................................. 64
4.2.2. Calibration results ................................................................................................................ 65
4.3. Evaluation of calibration ................................................................................................... 67
4.4. Groundwater balance ......................................................................................................... 68
4.5. Groundwater outputs (in water balance) ........................................................................... 69
4.5.1. Springs.................................................................................................................................. 69
4.5.2. Base flow to the rivers.......................................................................................................... 70
4.5.3. Abstraction of the wells........................................................................................................ 70
4.6. Model validation................................................................................................................ 70
4.7. Uncertainty of the model calibration and model limitations ............................................. 70
5. Akaki wellfield flow modelling.........................................................................................71
5.1. Introduction........................................................................................................................ 71
5.2. Hydrogeology of the wellfield........................................................................................... 71
5.2.1. Geology ................................................................................................................................ 71
5.2.2. Regional and local groundwater flow directions .................................................................. 73
5.2.3. Groundwater recharge and discharge conditions.................................................................. 73
5.2.4. Water level ........................................................................................................................... 74
5.3. Groundwater model of Akaki wellfield ............................................................................. 75
5.3.1. Spatial discretization of model grid...................................................................................... 75
5.3.2. Input parameters of the wellfield model domain .................................................................. 75
5.3.3. Model boundary conditions.................................................................................................. 76
5.3.4. Surface-groundwater interaction .......................................................................................... 77
5.4. Model calibration............................................................................................................... 78
5.5. Water budget of wellfield .................................................................................................. 83

iv
 5.6. Sensitivity analysis .............................................................................................................84
6. An advective solute transport modelling of Akaki wellfield, PMPATH .......................... 85
6.1. General................................................................................................................................85
6.2. Hypothesis and assumption of the solute transport (PMPATH) ........................................85
6.3. Factors governing contaminant transport ...........................................................................86
6.3.1. Implication of geology for transport .................................................................................... 86
6.3.2. Hydrogeological suitability for transport ............................................................................. 86
6.3.3. Geochemical processes affecting transport .......................................................................... 88
6.3.4. Effect of the slope on transport ............................................................................................ 88
6.4. Human activities having pollution potential in the area.....................................................88
6.5. Locations of potential contaminations................................................................................89
6.6. Predictive simulations of contaminant analysis .................................................................89
6.7. Pollutant travel time and direction of tracking computation ..............................................90
6.8. Implementation of groundwater protection zoning in Akaki wellfield ..............................94
6.9. Capture zone of the wellfield .............................................................................................95
6.10. Delineation of a 30 year protection zone of Akaki wellfield .............................................98
7. Conclusions, discussions and recommendations............................................................. 100
7.1. Discussion and conclusions ..............................................................................................100
7.2. Recommendations.............................................................................................................102
References ..........................................................................................................................................104

v
List of figures
Figure 1. Research methodology.......................................................................................................... 6
Figure 2. Research methodology for PMPATH................................................................................... 7
Figure 3. Location map of the study area............................................................................................. 9
Figure 4. Detailed location map of the Akaki wellfield..................................................................... 10
Figure 5. Location map of hydrological and meteorological stations in the Akaki catchment.......... 10
Figure.6. Mean monthly distribution of rainfall in Akaki catchment ................................................ 12
Figure 7. Variation of annual rainfall with elevation in Akaki river catchment ................................ 13
Figure 8. Spatial distribution of rainfall in the Akaki catchment....................................................... 14
Figure 9. Temporal variation of annual rainfall about the mean – Addis Ababa Observatory.......... 14
Figure 10. Relationship between temperature and rainfall in the catchment (1989 – 2005) .............. 15
Figure 11. Temperature variations in Akaki catchment ...................................................................... 15
Figure 12. Monthly variation of runoff at Akaki bridge, 1981 – 2005, (AAWSA, 2000) .................. 16
Figure 13. Simplified hydrogeological map (after AG consult, 2004) ............................................... 18
Figure 14. Drainage and urban development of Addis Ababa area .................................................... 20
Figure 15. Geomorphology of the Akaki catchment ........................................................................... 21
Figure 16. Landsat imagery with shaded relief map ........................................................................... 21
Figure 17. Graph of regression analysis of sum of anions against sum of cations ............................. 23
Figure 18. Graph of EC/100 (µs/cm) against sum of anions (meq/l).................................................. 23
Figure 19. Graph of EC/100 (µs/cm) against sum of Cations (meq/l) ................................................ 24
Figure 20. Piper Diagram showing the distribution of water types for water points .......................... 25
Figure 21. Stiff patterns of water samples from Akaki catchment...................................................... 25
Figure 22. North-South cross-section of well logs in the Akaki catchment (Dereje Nigusse, 2003) 27
Figure 23. Location map of the Akaki wellfield showing productive and monitoring wells.............. 28
Figure 24. Geological and geophysical log of Borehole 23 (BH 23) in Akaki wellfield.................... 30
Figure 25. Lithological cross – section along NW – SE in the Akaki wellfield (see Fig. 23) ............ 31
Figure 26. Lithological X- sections along E-W in the wellfield (2.2 km) (see Fig. 23) ..................... 31
Figure 27. Semi-log plot of time-drawdown curve for selected wells ................................................ 32
Figure 28. Akaki wellfield actual monthly production from July 2002 – October 2005.................... 33
Figure 29. Evolution of groundwater levels from May 01, 1999 to January 24, 2002 ....................... 34
Figure 30. Evolution of the piezometric levels during the pumping (1/11/2001-30/ 04/2002) .......... 35
Figure 31. Akaki wellfield water level monitoring report, 2002 to 2006, (see fig.23). ...................... 35
Figure 32. Hydrographs of MW1 (top) & TW1 (2005 - 2007 (bottom) ............................................. 36
Figure 33. N–S conceptual model of GW flow in the Akaki catchment (Demlie, 2007) ................... 37
Figure 34. 3D conceptual geospatial model frame work of Akaki catchment .................................... 40
Figure 35. Initial hydraulic head values of the catchment .................................................................. 42
Figure 36. Hydraulic conductivity values of the catchment................................................................ 43
Figure 37. Bottom of the aquifer......................................................................................................... 44
Figure 38. Top of the aquifer .............................................................................................................. 44
Figure 39. Vadose zone thickness map ............................................................................................... 45
Figure 40. Comparison of Specific capacity and values in the catchment......................................... 46
Figure 41. Comparison of pumping test and Logan estimated values in the catchment .................... 46
Figure 42. Transmissivity map of the area. ......................................................................................... 46

vi
Figure 43. 3D view of the Akaki catchment ........................................................................................50
Figure 44. Boundary conditions of the model domain.........................................................................50
Figure 45. Chloride map of groundwater in the Akaki catchment.......................................................52
Figure 46. Spatial distribution of groundwater recharge .....................................................................54
Figure 47. Baseflow simulation from total flow in Akaki catchment..................................................56
Figure 48. Schematic diagram of the water balance model .................................................................57
Figure 49. Monthly Flow Volumes at Aba Samuel .............................................................................59
Figure 50. Calculated total flow and base flow at Aba Samuel ...........................................................60
Figure 51. Relation between base flow and recharge with time ..........................................................60
Figure 52. Monthly rainfall (1985 – 1999) ..........................................................................................61
Figure 53. Groundwater surface interactions in the study area. ..........................................................62
Figure 54. Trial and error calibration procedures (Adapted from Anderson and Woessner, 1992) ....65
Figure 55. Scatter diagram of calculated and observed heads in the study area..................................65
Figure 56. Comparison between actually measured and model computed heads ................................66
Figure 57. Scatter diagram of calculated & observed heads (m), when 364 bore holes are used........67
Figure 58. Geological map of the study area (modified after AG consult, 2004)................................72
Figure 59. Dense fracture traces in the wellfield area (modified after Tamiru et al., 2005). ..............72
Figure 60. Comparison of ground surface elevation and normalized SWL in the wellfield. ..............74
Figure 61. Conceptual model of Akaki wellfield.................................................................................74
Figure 62. Block diagram shows conceptual model of Akaki wellfield ..............................................75
Figure 63. Input parameters of wellfield model...................................................................................76
Figure 64. Aba Samuel lake level (MOWR, 20007)............................................................................76
Figure 65. Boundary conditions of the wellfield model ......................................................................77
Figure 66. Wellfield steady state model calibration using static water levels with no abstractions....79
Figure 67. Calculated and. observed heads (m) of the wellfield without well abstraction..................79
Figure 68. Observed & simulated heads of Akaki wellfield, calibrating with pumping senario. ........81
Figure 69. Flow nets in the wellfield after calibrating the model with dynamic water level of 2004 .82
Figure 70. Hydraulic conductivity values (m/day) used for wellfield calibration ...............................82
Figure 71. Sensitivity plot of the calibrated model with respect to hydraulic conductivity ................84
Figure 72. Sensitivity plot of the calibrated model with respect to recharge ......................................84
Figure 73. Conceptualization of the process by which solutes transport by moving groundwater .....85
Figure 74. Rivers and groundwater connection; a gaining stream (left) and a losing stream (right). .87
Figure 75. Simulation of GW flow & solute transport (Chiang & Kinzelbach (1998)........................89
Figure 76. Various faces of an individual cell. ....................................................................................90
Figure 77. Flow nets showing column and row projections ................................................................91
Figure 78. Contaminant path lines of 100 years travel time, from pollutant sites ...............................91
Figure 79. Contaminant migration from pollutant sites with in 150 years. .........................................92
Figure 80. Contaminant entering to wellfield from kuye site at about 800 years................................93
Figure 81. Pollutant migration from Kality treatment plant for 300 years. .........................................93
Figure 82. Contamination capture (green) lines of wellfield for 30 years...........................................96
Figure 83. Capture zone of wellfield for 300 years .............................................................................97
Figure 84. A 30 year protection zone for boreholes in the wellfield ..................................................98
Figure 85. Protection zones delineated around the wellfield...............................................................99

vii
List of tables
Table 1. Digital elevation model versus mean annual rainfall of Akaki stations (1981 – 2003)....... 13
Table 2. Maximum and minimum monthly temperature in the catchment ........................................ 15
Table 3. ITC laboratory chemical analysis result (2008)................................................................... 22
Table 4. Ionic balance of the water samples ...................................................................................... 23
Table 5. Water type of the present study in the catchment (2008) .................................................... 24
Table 6. Summary of Phase I abstraction at Akaki Wellfield. (Nov 2001-Apr 2002) ....................... 33
Table 7. Annual production figures of the Akaki Wellfield (2002-2005) ......................................... 34
Table 8. Parameters considered in conceptualization of the system (Aquaterra consultant, 2000)... 38
Table 9. Major types of model boundary conditions(Aquaterra consultant, 2000) ........................... 49
Table 10. Summary of results of the monthly water balance model.................................................... 59
Table 11. Average monthly and annual recharge (mm) for each subcatchment. ................................. 60
Table 12. Model Calibration Performance Measures (Aquaterra consultant, 2000) ........................... 64
Table 13. Steady state flow Simulated vs. observed hydraulic heads (m.a.s.l).................................... 66
Table 14. Errors of the calibrated model.............................................................................................. 68
Table 15. Water balance in Steady state flow obtained through Akaki catchment model calibration 69
Table 16. Well abstractions applied to the model. ............................................................................... 78
Table 17. Errors of the calibrated model.............................................................................................. 79
Table 18. Boreholes used for model calibration under non-pumping scenario. .................................. 80
Table 19. Evaluation of calibration with pumping senario .................................................................. 81
Table 20. Observed and simulated heads of Akaki wellfield with pumping senario.......................... 81
Table 21. Groundwater balance with out well abstraction in the wellfield ......................................... 83
Table 22. Wellfield water balance with well abstraction, calibration by DWL(2004)........................ 83

viii
List of appendices
Appendix 1 Geologic logs, resistivity logs, and their corresponding values ....................................108
Appendix 2 Electric logging record (resistivity & self potential logging) .......................................112
Appendix 3 Hydrostratigraphic cross-section on the wellfield........................................................115
Appendix 4 E-W well log correlation of catchment (Dereje Nigusse, 2003) ...................................116
Appendix 5 How to correct ASTER DEM with topographic map ...................................................117
Appendix 6 Chloride concentration in Akaki catchment (2007 – 2008) ..........................................119
Appendix 7 Plots of time-drawdown curve for selected bore holes .................................................120
Appendix 8 Semi variogram models for aquifer parameter determination.......................................121
Appendix 9 Some of model input parameters to Akaki catchment model........................................122
Appendix 10 Wells, head dependent boundaries, & rivers applied to wellfield.................................125
Appendix 11 Akaki wellfield monitoring wells..................................................................................131
Appendix 12 Subcatchments and soil permeability of Akaki area .....................................................132
Appendix 13 Monthly rainfall data for the water balance model .......................................................134
Appendix 14 Pictures of the area ........................................................................................................136

ix
Acronyms
AAWSA: Addis Ababa Water and Sewerage Authority
AESL: Associated Engineers Service Limited
COMPLANT: Chain National Complete Plant Import and Export Corporation
WWDE: Water Well Drilling Enterprise
EPA: Environmental Protection Authority
EMA: Ethiopian Metrological Agency
BCS: Black Cotton Soils
DEM: Digital Elevation Model
GPS: Global Positioning Systems
CEC: Cation Exchange Capacity
NE, NW, SE, SW: North-East, North-West, South-East, South-West
E-W, N-S: East-West, North-South
UTM: Universal Transverse Mercator
EHA: Ethiopian Highway Authority
GWL: Groundwater Level
SWL: Static Water Level
DWL: Dynamic water Level
TDS: Total Dissolved Solids
µS Micro Siemens
mm Milimeter
masl Meter above mean sea level
Fig. Figure
GIS Geographic Information System
EC Electrical conductivity
CMB Chloride mass balance
AAO Addis Ababa observatory
ITCZ Inter-tropical convergence zone
K Hydraulic conductivity
T Transmissivity
MCM million cubic meters
mg/l Milligram per liter
mm millimetre
ppm parts per million
SC Specific capacity
MER Main Ethiopian Rift
GW Groundwater

x
1. Introduction

1.1. General
Rapid population growth, urbanization and increased demand of water in domestic and industrial
production have led to fresh water shortage in many parts of Ethiopia and to an increasing dependence on
groundwater. With continuous increased withdrawals from groundwater reservoirs results in systematic or
continuous lowering of water table. Hence, the need for better management of water resources is crucial.
This task becomes even more pressing as urbanization, industrialization and development advances. Easily
fragile resource like water can be affected by pollution. Groundwater modelling is an essential tool to
evaluate the groundwater flow and quantifying its potential. It also helps understanding and predicting the
behaviour of the groundwater system in response to future stresses due to abstraction or meteorological
forcing. A mathematical model simulates groundwater flow indirectly by means of governing equations
which represent the physical process that occur in the system together, with equations that describe the
head or flow along the boundaries of the model (Anderson and Woessner 1992). The same points apply to
groundwater flow and contaminant transport modelling. Tamiru (2001) indicated that there is surface water
pollution as a consequence of industrial effluents, municipal and house hold waste disposal in the area of
investigation. Besides that, Gizaw (2002) has also conducted hydrochemical and environmental
investigation of the Addis Ababa region which can serve as information for this research.

In the region rocks are exposed to tectonic effects of the Ethiopian rift with plenty of faults, lineaments,
fissures, fractures, conduits and joints trending NE-SW, E-W, N-S, and NW-SE. Moreover the density of
the structures increases to the SE of Addis Ababa around the Akaki wellfield. Soil investigation done in the
area also reveals a relatively higher hydraulic conductivity of most soil types. Most pollutants irrespective
of source are transported overland and through the soil by rain water and ultimately find their way into the
groundwater. The soil filters the water, absorbs and removes many contaminants, though the excess
contaminants may pass through the zone of aeration and enter the groundwater in the zone of saturation.

The delineation of a protection zone in Akaki wellfield is the process that determines the geographical area
that should be included in a protection zone program. This area of land is managed to minimize the
potential of groundwater contamination by human activities that occur on the land surface or in the
subsurface. Proper implementation of aquifer protection zoning will ensure water quality benefits in the
long term. As a result, the unpolluted water sources will aid in good health of the people, animals and
ecosystems. Therefore, delineating capture zones and wellhead protection areas; prediction of the
contaminant distribution in time and space is necessary for managing ground water resources and designing
optimal mitigation plan in areas like the Akaki wellfield, which is at risk due to unregulated solid and
liquid waste disposals from its upper stream areas. For this purpose, an advective transport modelling
known as PMPATH will be applied in the area under investigation. PMPATH will also be used to analyze
flow paths and travel times in the groundwater.

1
1.2. Statement of the problem
The Akaki wellfield is the largest wellfield in Addis Ababa city and is serving as source of water supply for
the city. The groundwater table of the wellfield is continuously declining due to abstraction of large
amount of water mainly for water supply of the city. In order to better understand and to predict the flow
system, groundwater flow modelling needs to be applied to regional groundwater system and the Akaki
wellfield in particular.

The increased need for drinking water calls also for careful consideration and integration in the
development process of all environmental factors. There is no detailed national investigation, which clearly
puts and determines how contaminants (pollutants) migrate in subsurface environments of the proposed
area, although the continual disposal of unknown amount of sewage, garbage, and even toxic pollutants
into Akaki river and its tributaries is clearly observed. Such pollutants may eventually enter the aquifer
system through porous, permeable media that are highly dissected by numerous structures. This may pose
not only a problem in utilizing the resource as drinking water but also incurs a huge later investment to
clean it, or even impossible to pump the polluted aquifer if once it has been affected by such pollutants.
Since the contaminants that reach the groundwater generally move very slowly, continued leakage at one
site will lead to a gradual accumulation. In most natural settings, pollutant accumulations in the
environment are not very serious because the natural concentrations of these contaminants are low in
waters and soils (Gizaw, 2002). The problem aggravates when human activities locally upset the natural
cycle. Cities and other residential communities contribute mostly sewage, with traces of household
chemicals mixed in. Most industries and factories dispose their effluent through fall pipes into the
environment increasing the variety of pollutants in water resources. Therefore, modelling the transport of
contaminants in a wellfield, which is clearly under threat by industrial wastes, is not only a timely venture
but also a strong instrument in alleviating relevant problems of drinking water in a city where the
population is increasing at an alarming rate. Environmentally incompatible industries like skin and hide,
chemical, metal and textile factories are unfavourably located along Akaki road. The NO3- detected covers
a wide range (0.04-241 mg/l); the Mn2+ level reached up to 1.5 mg/l; Cd2+ in EP-6 well were 19.74 ȝg/l;
and Cr3+ was 182 ȝg/l in Tiliku Akaki river sampled at Akaki bridge (Gizaw, 2002). All of them exceed
their respective WHO guideline limit of 50 mg/l, 0.1 mg/l, and 3ȝg/l respectively. The amounts are more
likely to be originated from industrial activities.

1.3. Objectives

1.3.1 General objective

The general objective of this study is to investigate the groundwater flow system in the Akaki catchment as
well as its wellfield by using numeric groundwater modelling. Besides that travel times and paths of
pollutant migration in wellfield will be assessed.

1.3.2 Specific objectives

The primary objectives will be achieved through the following sub objectives:
ƒ Analyze hydrochemical data using geostatistics and GIS to determine the recharge, and its spatial
distribution.
ƒ Develop conceptual models and based on that, to build a numeric model, which can be used to
simulate the groundwater flow under steady state conditions.

2
 ƒ Generate hydraulic heads in the wellfield by numerical groundwater flow model and calculate
velocity distributions;
ƒ Determine travel time and travel paths of contaminants in the wellfield which can help in
delineating the wellhead protection area (capture zone).

1.4. Research questions

By these objectives the following research questions are formulated:
ƒ Can hydrochemical data be used to evaluate the recharge?
ƒ Can a conceptual model be transformed in a numeric model that represents the simplified field
situation of the aquifer system?
ƒ Can a calibrated groundwater flow model improve our understanding of groundwater flow and
contaminant flow pattern in the wellfield?
ƒ How is the abstraction rate from the wellfield affecting the contaminant movement?

1.5. Research hypothesis

ƒ The calibrated groundwater flow model of Akaki wellfield can simulate observed heads with less
than 15% (0.15 m) of root mean squared error.
ƒ An advective contaminant transport model known as PMPATH can simulate contaminant
movement in the area with time which can be used for delineation of protection zone.

1.6. Previous works

Concerning geology of Addis Ababa region, different papers have been written. Hydrogeological study of
Akaki catchment with special emphasis on the problems of groundwater recharge has been carried out by
Demlie (2007). Previously detailed hydrogeological investigation has been also carried out by Anteneh
(1994) in his masters thesis “Hydrogeology of Akaki Area”. As cited in a thesis, many geoscientists like
Mohr (1967), Morton et al. (1966), Kazmin (1978), Morton et al. (1979), Zanettin and Justin Visitin
(1974), Mohr .P.A.(1983) have discussed the geology of Addis Ababa region in their studies. Girmay and
Assefa (1989) discussed the volcanic strategraphy of rock units outcropping at the vicinity of Addis Ababa-
Nazareth area. Regarding surface and ground water potential of Addis Ababa region, AAWSA and
SEURECA (1989) have conducted valuable works. Assessments to assure the feasibility of surface and
groundwater as a source of water supply in Akaki area have been carried out by AWSA (2000). The
geology and hydrogeology of Akaki (Sekelo) sub basin with particular emphasis to its hydrochemistry and
interaction between the surface and ground water has been investigated by Aynalem (1999) in his master
thesis “Water Quality and Ground Water/Akaki River Interaction in the Sekelo Basin (Lower Akaki River
Sub Basin)”. Specific references to water well drilling, Construction, Pumping test data, water quality
data, Geological and geophysical logs in Akaki wellfield are included in water well construction report by
AAWSA as a client and WWDE (Water Well Drilling Enterprise) as a contractor in 1996. Similar activities
have been carried out for 25 water wells drilled in the Akaki wellfield for the supply of Addis Ababa by
AAWSA as a client and COMPLANT as contractor in 1997.

In Addis Ababa, and its surrounding areas, both surface and groundwater resources have been investigated
in terms of potential, and vulnerability by a relatively good number of investigators such as Vernier (1993);
AAWSA-SEURECA (1991); Anteneh (1994); Eccleston (1997); Aynalem (1999); AAWSA (2000);
Gebrekidan (2000); Tamiru (2001); and Gizaw (2002). EPA (1997) has surveyed pollutant load on three

3
rivers of the Addis Ababa city. These studies, although they vary in scope and degree of geological and
geochemical information, have stressed that the quality of surface water is often affected by uncontrolled
waste disposal of domestic and municipal wastes and industrial effluents. They further indicated that these
would have potential impact on the quality of groundwater of the region. Demelie (2007) studied the
recharge in the area and he stated that recharge takes place over the entire surface area, major recharge
takes place within the Intoto sector of the catchment, serving as a so-called mountain block recharge.
Moreover, hydrochemical and environmental isotope data indicated additional recharge sources from
wastewater, leakage from mains and reservoirs. In addition to that, Tamiru et al. (2005) did aquifer
vulnerability of the whole Akaki catchment and water quality assessment.

1.7. Methodology
The study is conducted using data issued from literature; secondary data collected from offices, water
sampling and laboratory analyses. The methods to be followed in this research are based on the objectives
formulated in section 1.3.

1.7.1. Pre fieldwork

Literature review and collection of satellite images such as ASTER and LANDSAT were the core methods
that were employed in the office. The literature review was done based on both national and international
literatures. Collection of many secondary data as much as possible, preparing satellite imageries, and
becoming familiar with various appropriate Softwares like PMPATH, Aqua Chem. 4.0, Aquifer test,
Golden Surfer Software 8, Arc View 3.3, Visual Modflow pro, Arc GIS 9.2 etc as well as scientific
instruments such as, pH meters, GPS, EC meter and sample bottles were some of pre field works
accomplished. In addition to this, literature review related to recharge estimation and principle of ground
water and contaminant flow modelling was part of the pre fieldwork. The reports of the area were reviewed
to get insight about the problem and to define the work direction of the research. Besides that field
equipment for fieldwork such as GPS, pH meter, altimeter, compass, dipper and EC meter were prepared.

1.7.2. Field work

The following points summarize the field work:
ƒ Collection of water samples from springs, boreholes, rainfall and surface water bodies for chemical
analysis (major cations and anions). Samples were filtered by 0.45µm or less before actually
sampled and prepared for preservation. they are taken in polyethylene bottles with tight cap and
then stored in relatively cold areas to prevent reaction of constituents of water in the sample. In
addition to this, water samples were acidified using HCl or HNO3 for storage purpose depending
on the purpose of sampling. The analysis result from the water samples was applied for
identification of the regional and local flow systems, determination of water types and for
assessment of groundwater quality in the area.
ƒ Spot measurements of EC, pH, bicarbonate (alkalinity), Cl, and NO3- were carried out.
ƒ Meteorological data were collected from National Meteorological Service (Rainfall, temperature,
wind speed, humidity, etc).
ƒ Taking readings of boreholes location and elevation using GPS
ƒ Collection of monitoring groundwater level data from Addis Ababa Water and Sewerage Authority
and related organizations.

4
 ƒ Well completion and pumping test data are also crucial requirement for characterization of the
aquifer and are collected from Water Works Enterprise, Addis Ababa water and sewerage authority
(AAWSA) and other related organizations.
ƒ Collecting of geophysical data conducted during the feasibility studies (available reports).
ƒ Delineation and field verification of the study area and identification of physical boundaries
ƒ Locating pollutant sources or sites.

1.7.3. Post field work (data processing and analysis)

During this stage, the data collected during the pre field and fieldwork was processed and analyzed.
ƒ For the processing purpose, GIS and geostatistics were extensively used. For example, ILWIS was
applied for kriging in order to interpolate point measurements.
ƒ Well completion data analysis to determine the position of screen, aquifer thickness and number of
layers in combination with well log cross- section.
ƒ Determination of aquifer parameter maps
ƒ The water samples collected during the fieldwork were analyzed and processed for the major
cations and anions to understand the geochemical property and water type detection. The
laboratory analysis result were processed and presented using AQUACHEM software. The
determination of the chloride concenteration in rainwater and its concentration in ground water was
part of this analysis as it will be used for recharge estimation by using chloride mass balance
method.
ƒ Constructing a steady state groundwater flow models and calibration of the models.
ƒ Simulation of the migration of pollutants, analysis of the intrusion of contaminant plume, and
prediction of the time and direction (flow path) of contaminant in fractured rock systems will be
carried besides laboratory work.

1.7.4. Methods to apply PMPATH (Advective transport model)

PMPATH is powerful tool for wellfield management and decision making because of this, it were used for
the following purposes:
ƒ calculating the travel times of contamination from their sources to wellfield and path lines of the
contamination
ƒ Delineation of capture zones of pumping wells and wellfield protection areas
General assumptions used when using PMPATH includes:
ƒ Fluid properties are homogenous,
ƒ Concentration changes do not affect the fluid density and viscosity and hence fluid velocity
ƒ Contaminant moves with the same speed as ground water
ƒ Pollutant movements are dependent specifically to effective porosity.

5
Figure 1. Research methodology for flow modeling

6
Figure 2. Research methodology for PMPATH

7
1.8. Thesis outline

The research is sub divided in to seven chapters and the contents are summarized briefly as:

Chapter1: discuses about introduction of the research that include the problem statement, the objective of
the research and research questions which the research tries to answer on the basis of the available data.
Methods Applied and previous works are also discussed in this introductory chapter.

Chapter 2: Gives an introduction of study area in terms of location, climate, topography, hydrology,
geology, hydrogeology, hydrochemistry, well logs and subsurface conceptualization, aquifer types and
water level monitoring in response to external stresses.

Chapter 3: Numerical groundwater modelling mainly designed to discuss the model setup, code selection,
spatial discritization, and input parameter determination. This chapter also discusses the recharge in Akaki
catchment.

Chapter4: Describes the Akaki catchment’s steady state model execution, calibration and water balance
assessment. This chapter also addresses model limitations and sensitivity analysis.

Chapter 5: Describes characteristics of the wellfield, development of its conceptual model, input
parameters, model calibration and water balance calculations.

Chapter 6: Describes an advective solute transport modelling on the wellfield particularly giving focus on
factors affecting contaminant transport, human activities having pollution potential, delineation of capture
zone and wellhead protection areas of the wellfield. In addition, this chapter explains the distribution of
contaminant in groundwater in terms of space and time.

Chapter 7: Discussion, Conclusion and recommendations

Discusses the result of the recharge estimated by different methods, subsurface conceptualization, aquifer
types and travel paths and travel time of contaminants in the wellfield. In this final chapter, matters which
can not be addressed are also recommended to be implemented for the future.

8
2. The study area

2.1. Location, digital elevation model and climate

The Akaki catchment is located in central Ethiopia along the western margin of the main Ethiopian rift
valley. The catchment is situated at the north western Awash River basin between 8046’–9014’ N and
38034’–39004’E. It is bounded to the north by the Intoto ridge system, to the west by Mt. Menagesha and the
Wechecha volcanic range, to the southwest by Mt Furi, to the south by Mt Bilbilo and Mt Guji, to the
southeast by the Gara Bushu hills and to the east by the Mt Yerer volcanic centre. The Akaki catchment has
an area of about 1500 km2. Addis Ababa is located at the centre of the catchment (Figure 3). Surface water
reservoirs located within the study area include Legedadi, Gefersa, Dire and Aba Samuel (See figure 5).

Figure 3. Location map of the study area

9
Figure 4. Detailed location map of the Akaki wellfield.

(m)

(m)

Figure 5. Location map of hydrological and meteorological stations in the Akaki catchment

10
2.1.1. Digital elevation model
A digital elevation model, DEM is extracted from Level 1A image of ASTER using ERDAS IMAGINE to
a 15 meters resolution. Ground control points were prepared from the topographic maps of the study area to
check the vertical accuracy of the DEM. The vertical accuracy of the DEM was adapted to the topographic
map’s vertical accuracy by extracting elevation value from topographic map at a number of check points.
The procedure is described in detail in Appendix 5. Corrected Elevation from the ASTER DEM is applied
to define the ground elevation of the boreholes, to define the aquifer top and bottom, to construct cross-
sections, to determine rainfall distribution as well as recharge in the catchment, and to define elevation of
the bottom of the riverbed for the use of river package. Furthermore the DEM plays a key role in defining
of the model boundary in the conceptual model formulation.

2.1.2. Climate
Despite its proximity to the equator, the study area experiences a temperate Afro-Alpine climate. Daily
average temperatures range from 9.9 to 24.6 0C (Fig.11) and annual mean rainfall is 1224 mm/yr, as
measured at the catchment by hyposometric method (see Fig.8, rainfall map). The climate of the Akaki
catchment is characterized by two distinct seasonal weather patterns. The main wet season, locally known
as Kiremt, extends from June to September, contributing about 70% of the total annual rainfall (Figure 6).
A minor rainy season, locally known as Belg, contributes moisture to the region from mid February to mid
April (Daniel, 1977).

EMA (1988) defined five traditional climatic zones: "Kur" (Alpine), 3000m and above; "Dega"
(temperate), 2300m to about 3000m; "Weina Dega" (Sub tropical), 1500 to about 2300m; "Kolla"
(Tropical), 800m to about 1500m and "Bereha" (Desert), less than 800m. The Climate of Addis Ababa is
Woina Dega (Daniel (1977). The Rainfall of the area is nearly bimodal (two peaks): the Belg rains
(February to May) and Kiremt (main) rains (June to September). The highest rainfall peak is in August.

2.1.3. Precipitation
The variation in the seasonal distribution of rainfall in Ethiopia can be attributed by the reference to the
position of the Inter-Tropical Convergence Zone (ITCZ), the relationship of between upper and lower
air circulation, the effects of topography and the role of local convection currents and the amount of
rainfall (Daniel, 1977).
According to Daniel (1977) classification of Ethiopia's rainfall region, Addis Ababa is located in the
region where the rainy months are contiguously distributed (Regime IE). In this region there are seven
rainy months from March to September/and the small rains occur from March to May. The big rains are
from June to September. High concentration of rainfall occurs in July and very high concentration in
August.
As stated above, prior to determine rainfall distribution in the catchment, the ASTER DEM was
corrected from vertical scale errors by adapting to topographic map accuracy. After obtaining corrected
digital elevation model for the whole catchment, the elevations of rainfall stations are extracted from
digital elevation model surface and mean annual rainfall of each station is determined. Finally the
elevations and mean annual rainfall of all stations are correlated to determine the spatial rainfall
distribution of the catchment.

11
 400 400
400
400

350 350
350
350

300 300
300
300
Rainfall (mm)

(mm)
Rainfall (mm)
250

Rainfall (mm)
250
250
250
200 200
200
200

Rainfall
150 150
150
150
100 100
100
100
50 50
5050
0
0
00
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Jan Feb
Feb Mar
Mar Apr
Apr May
May Jun
Jun Jul
Jul Aug
Aug Sep
Sep Oct
Oct Nov
Nov
Nov Dec
Dec
Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct
Month Month
Month
Month

Rainfall distribution at Sendafa station (SEN) Rainfall distribution At Akaki station (AK)

400 400
400
400
350 350
350
350
300 300
300
300
m ))
Rainfall (mm)

)
(m m

250
250
(m m

250 250
Rainfall (m

200
200
Rainfall

200 200
Rainfall

150 150
150
150

100 100
100
100

50 50
50
50

0 000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Jan Feb
Jan Feb
Feb Mar
Mar
Mar Apr
Apr
Apr May
May
May Jun
Jun
Jun Jul
Jul
Jul Aug
Aug
Aug Sep
Sep
Sep Oct
Oct
Oct Nov
Nov
Nov Dec
Dec
Dec

Month Month
Month
Month

Mean monthly rainfall distribution At Debreziet station (DZ) Mean monthly rainfall distribution At Entoto station (INT)

400
400
350
350
300
300
Rainfall (mm)

250
Rainfall (mm)

250

200 200

150 150

100 100

50 50

0 0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month Month

Mean monthly rainfall distribution At Addis Ababa Mean monthly rainfall distribution At Addis Ababa
university Geophysical observatory station (AAO)
Bole station (AAB)

400

350

300
Rainfall (mm)

250

200

150

100

50

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month

Mean monthly rainfall distribution At Melka Kuntur

station (MK)

Figure.6. Mean monthly distribution of rainfall in Akaki catchment

12
Table 1. Digital elevation model versus mean annual rainfall of Akaki stations (1981 – 2003)
Name of stations X Y DEM elevation RF
Akaki 476486 979917 2040 1117.1
Sendafa 502257 1011809 2532 1286.6
Debreziet 495194 966825 1850 900
AA0 471986 997321 2457 1244.6
Melka Konture 456328 962787 1880 947.6
Bole 476413 992272 2307 1178.7
Intoto 474400 1004200 2772 1394.3

As shown in the Table 1 and Figure 7 below, the annual rainfall value is observed to increase with
elevation. Therefore, the hypsometric method is used for rainfall estimation in the catchment.
Therefore, the mean annual rainfall is 1224 mm/yr with maximum and minimum RF of 1705 and 970
mm/yr. Using hypsometric relationship between rainfall and elevation, the rainfall distribution map of
Figure 8 is produced using ILWIS script.

1500
Rainfall (mm)

y = 0.4998x + 21.956
1400 2
R = 0.957
1300
1200
1100
1000
900
800
1500 1700 1900 2100 2300 2500 2700 2900
Elevation (m)
Figure 7. Variation of annual rainfall with elevation in Akaki river catchment

13
 Figure 8. Spatial distribution of rainfall in the Akaki catchment

2000

1800
Rainfall (mm)

1600

1400

1200

1000

800
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Year
Figure 9. Temporal variation of annual rainfall about the mean – Addis Ababa Observatory

2.1.4. Temperature
Under normal conditions, air temperature decrease with increasing altitude at a mean rate of 0.7 oC for
every 328 feet (Fetcher, 1998). This works also in Ethiopia where temperature decreases with increasing
elevations. Daily average temperatures range from 9.9 to 24.6 °C. and mean monthly temperature ranges
between 15°C and 18 °C.

14
Figure 10. Relationship between temperature and rainfall in the catchment (1989 – 2005)

Table 2. Maximum and minimum monthly temperature in the catchment

Temp. Jan. Feb. Mar. Apr. May June July Aug. Sep. Oct. Nov. Dec. Ann. Ave.
Max. 23.3 24.1 24.6 23.9 24.6 22.9 20.3 20.1 21.1 22.4 22.6 22.8 22.7
Min. 8.2 9.5 10.9 11.5 11.7 10.8 10.8 10.8 10.5 9.2 7.9 7.5 9.9

30
25
Temprature(oC)

20
Max.
15 Min.
10
5
0
Apr.

Aug.
May

July
Feb.

Mar.

Sep.

Nov.
June
Jan.

Oct.

Dec.

Months
Figure 11. Temperature variations in Akaki catchment

2.2. Hydrology
The Akaki River, left bank tributary of the Awash River, has an extensive drainage system with a
catchment area of 1500 km2. It originates from the Intoto range, north of Addis Ababa, and drops about
600 m in a river length of 95 km from its origin to its confluence with the Awash River near Dodota. The
Akaki river watershed includes two main river systems: the Akaki proper (big Akaki river), draining the
eastern part of the catchment area, and the Little Akaki, draining the western portion. These two rivers
meet in Abba Samuel reservoir, a man-made lake formed by damming of the Akaki River in the 1930's.
South of the Abba Samuel dam, the Akaki river has cut a gorge up to 100 m deep, which extends for about
8 km. The Akaki river flows into the Awash river at a point approximately 18 km to south of the Abba
Samuel reservoir. The Akaki wellfield is situated within the drainage basin of Sakelo River, which flows
into Akaki River just before Abba Samuel reservoir.

15
The Akaki River is gauged since 1981 at the Addis Ababa - Debre Zeit Road Bridge, with a catchment area
of 885 km2. This station does not include the flow from the little Akaki River. This hydrometric station is
equipped with an automatic water level recorder and 7 staff gauges (1- m long each). The mean annual
discharge of Akaki river at this point is computed from the record for the period of 1981 to 2005, and is
found to be 9517 l/s or 10.75 l/s/km2 which is equivalent to mean annual yield of 339 mm. The monthly
distribution of runoff volume at Akaki Bridge in an average year is shown in Figure 12.

140
120
Runoff (MCM)

100
80
60
40
20
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Month
Figure 12. Monthly variation of runoff at Akaki bridge, 1981 – 2005, (AAWSA, 2000)

As (AAWSA, 2000) indicated that more than 82% of the annual runoff is generated in July, August and
September only; emphasizing the fact that groundwater contribution to surface water is low as compared to
direct runoff generated during the main rainy season. A trend of increase in low flow with time has been
observed in the flow record. This may be attributed to the ever-increasing wastewater flow into Akaki
River as Addis Ababa is expanding in terms of development and population. A significant portion of the
low flow at this station is, in fact, sewage originated.

There are three man-made reservoirs in the Akaki catchment, commanding a sizable part of the catchment
area. They facilitate transfer of water from one part of the catchment to the other in the form of water
supply for domestic and industrial consumptions. One has to consider this human influence on the
hydrological process while studying the water balance of the catchment. It is also worth mentioning that
sizeable quantity of sewage, originating from Addis Ababa City, is discharged into both Akaki Proper and
Little Akaki rivers. In particular, the sewage flow has modified the low flow regime of the river to a great
extent, and it could pose a serious threat to the groundwater in the area in terms of pollution if the former
recharges the latter.

2.3. Geological and hydrogeological setting

Owing to its location along the western margin of the main Ethiopian rift, the geological history of the
Akaki catchment is an integral part of the evolution and development of the Ethiopian Plateau and the rift
system. The catchment is covered by volcanic rocks overlain by fluvial and residual soils, in which black
cotton soils are predominant, varying in thickness from a few centimetres to about 20 m (AAWSA, 2000).
The main lithologies include basalts, rhyolites, trachytes, scoria, trachy-basalts, ignimbrites and tuff
(Figure 13). These highly weathered, fractured lithologies favour the circulation and storage of subsurface
water.

16
2.3.1. Geology
Several geological studies have been carried out in the region. The most important ones are Morton et al.
(1979) and Girmay and Assefa (1979). From field investigations and previous geological studies a
simplified hydrogeological map was established (Fig. 13). According to AAWSA-Seureca (1991) the
following lithostratigraphic units can be identified in the catchment.

Alaji Formation:
It includes rhyolites, trachytes, tuff, agglomerate, and aphianitic basalt. This unit is dominant in the
northern and central part. Earlier works further subdivided this series into Alaji rhyolites and Intoto Silicics
(Zanettin and Justin- Visentin, 1974). It extends from Intoto to the northern adjacent Blue Nile (Abay)
basin (Girmay, and Assefa, 1989). The Intoto trachyte overlies the Alaji basalts. The Intoto Silicics
represents massive Oligocene fissure-basalt, rhyolites, and trachytes with minor welded tuff and obsidian
(Morton et al., 1979).

Addis Ababa Basalts:

These basalts overlay the Intoto Silicics and cover the central and southern part of Addis Ababa. Usually
individual flows are easily observed and paleosoils and scoraceous horizons are common in many places.
Olivine porphyritic basalt outcrop in central Addis Ababa with a thickness varying from 1 m or less in the
foothills of Intoto to more than 130 meters in central Addis Ababa (Vernier, 1985).

Younger Volcanics
This group includes the Nazareth Group and Bishoftu Formations.
The Nazareth Group rocks outcrop dominantly south of the Filwuha Fault. They are composed of
aphianitic basalts, welded tuffs, ignimbrites, trachytes, and rhyolites. It is overlain by aphianitic basalt and
underlain by porphyritic basalt. The exposed thickness of the lava sequence is about 500 m (Girma, 1994).
The Bishoftu Formation consists of olivine porphyritic basalt, scoria, vesicular and scoraceous basalt, and
locally trachy-basalt lava flows. They are localized in the south and are 20 to 40 m thick in the Akaki
wellfield. Locally it is overlain by scoria, tuff, sand, and gravel. This unit forms the major aquifer of the
region.

Recent Deposits
These include alluvial, residual, and lacustrine deposits. The thickness varies between 5 m and 50 m near
river banks in the south (AAWSA, 2000). It is often overlain by dark younger black cotton clayey soils.
Alluvial deposits are found in some places along the Little- and Big-Akaki rivers, particularly south and
southwest of Addis Ababa. Residual soils are located in the central, southeast, northeast, and western
flatplains.

2.3.2. Hydrogeology
The groundwater circulation and the dispersion of pollutants depend on the hydrogeological characteristics
of the material more specifically hydraulic properties such as porosity, permeability and . The origin, flow
and chemical constituent of groundwater is controlled by the type of lithology, distribution, thickness and
structure of hydrogeological units through which it moves. Moreover, the stresses due to tectonism and
weathering govern the hydrogeochemical characteristics of earth materials. Volcanic rocks mainly basalts,
rhyolites, trachytes, scoria, trachy-basalts, welded and unwelded tuffs are the dominant rock outcrops in the
area. Besides, unconsolidated materials of different origin also occur in the study area. These rocks are the
major groundwater supply for large parts of Addis Ababa.
17
The main structures, joints, fractures, and normal faults, are all related to the extensional rift tectonics in
the area. A prime example is the Filwoha Fault, along which thermal activity is observed. The aquifer
properties in the Akaki catchment are controlled by the litho-stratigraphy of the volcanic rocks and the
structures that affect them. More specifically, the hydraulic complexity of these volcanic rocks is caused by
their complex spatial distribution, their different reciprocal stratigraphic relationships, their significant
compositional, structural and textural variability, and their different levels of tectonization and weathering
(Vernier, 1993). These volcanic aquifers can be considered as a double porosity medium due to the fact
that both the matrix and the fracture porosity contribute to the transmission and storage of subsurface
water.
ƒ Scoria, scoraceous basalt and intra-formational gravel and sand layers constitute highly productive
aquifers with primary porosity and permeability.
ƒ Highly weathered and fractured basalts, fractured tuff, ignimbrite and other pyroclastics constitute
highly productive aquifers of secondary porosity and permeability.
ƒ Basalt with some fractures, vesicles and sparsely spaced joints, ignimbrites and agglomerates form
moderately productive aquifers in the area (Girma, 1994; AAWSA, 2000; Tamiru, 2001).
These units have been grouped into an uppermost shallow phreatic aquifer constituted by layers of
alluvial sediments, weathered and fractured volcanics and a confined–semi-confined volcanic aquifer of
widespread areal coverage (Girma, 1994). These multilayered, heterogeneous unconfined, semi-confined
and confined aquifers feature multi-stock works. Furthermore, it was observed that highly transmissive
areas do not correspond to high storativity areas, because of differential fracturing and weathering of the
volcanic aquifers (AAWSA, 2000).

Figure 13. Simplified hydrogeological map (after AG consult, 2004)

Relatively better hydrogeological information is collected in the central and southern part of the catchment
where most of the wells are concentrated. Major contributions to the hydrogeology of the area come from
studies related to the water supply of the city of Addis Ababa (AAWSA-Seureca 1991; AAWSA 2000) and
academic works (Girma, 1994; Aynalem, 1999; Gizaw, 2002; Demlie, 2007). These studies provided

18
important information to develop the conceptual hydrogeological model of the catchment. Field
hydrogeological investigation indicates variable aquifer hydraulic parameters. Most of the aquifers are
unconfined with productivity increasing towards the south.

The borehole logs show variable lithology and degrees of fracturing. However, due to the high permeability
the different hydrostratigraphic units are hydraulically connected. The pumping test data indicates that the
system behaves as an unconfined aquifer. The existing data shows that the catchment is composed of both
inter-granular and fracture permeability aquifers. The inter-granular porosity aquifers are alluvial sediments
and pyroclastic and scoraceaous volcanic rocks forming the upper more permeable layer. The fractured
aquifers are composed of basalts, ignimbrites, trachytes, and rhyolites with a relatively less permeable
bottom second aquifer system. Basalts and ignimbrites are generally highly fractured and permeable and
have been classified as good aquifers. The scoraceous and vesicular basalts have yields up to 1296 m3/day.
Faulted ignimbrites and basalts make good aquifers, while less fractured and less jointed basalts, including
alkaline flood basalts of the Addis Ababa area, form moderate aquifers with yields of up to 432 m3/day and
260 m3/day, respectively. Poor aquifers are fine-grained alluvial deposits with volcanic ash and lacustrine
deposits. In contrast to this general observation, the alluvial deposits of sand and gravel types have a very
good yield forming the shallow aquifers. Based on the analysis of existing well data, as well as those
gathered for this research, confined aquifers have been identified in the city of Addis Ababa. These are
mainly hot waters along the Filwoha fault, which are considered to be isolated and which have no direct
connection with the unconfined system.
Therefore, according to previous studies, the main aquifers in the Akaki catchment can be categorized into
three groups:
1. Shallow aquifer: made of weathered volcanic rocks and alluvial sediments along the river valleys.
2. Deep aquifers: made of fractured volcanic rocks at which boreholes are drilled for drinking water
supply purpose.
3. Thermal aquifer: that is located at depth greater than 300m.

2.4. Geomorphology and drainage system

Addis Ababa is located on the shoulder of the Western Main Ethiopian Rift Escarpment. The morphology
is a direct reflection of the different volcanic stratigraphic successions, tectonic activities and the action of
erosion between successive lava flows.
The city was founded at the southern flank of Entoto ridge (3565 m a.s.l.) and expanded in all directions.
This ridge marks the northern boundary of the city following the east-west trending major fault (Ambo-
Kassam). Other prominent volcanic features surrounding the city are Mt. Wochacha in the west (3385m
a.s.l.), Mt. Furi (2839m a.s.l.) in the southwest and Mt. Yerer (3100 a.s.l.) in the southeast.
These typical volcanic features are mainly built up of acidic and intermediate lava flows. Thus, they are
characterized by rugged landscapes and steeper slopes. The general inclination of the slope becomes lower
towards the southern part of the project area particularily around the wellfield.
The center of the city lies on an undulating topography with some flat land areas. The topography is
undulating and form plateau in the northern, western and southwestern parts of the city, while gentle
morphology and flat land areas characterize the southern and southeastern parts of the city. Moreover, it is
not uncommon to see sharp changes in the inclination of the slope and some flat land areas in different
parts of the city.

19
The Akaki River catchment, which includes the city of Addis Ababa and the wellfield area, is an extensive
drainage system arising from the Entoto Mountain Range north of Addis Ababa. The south eastern drainage
divide, which separates the Akaki and Dukem River drainage systems, passes very close to the Akaki
wellfield. This divide includes peaks of Gara Bushu (2346 m), Mt. Guji (2475 m), and Mt. Bilbilo (2380
m).The Akaki wellfield is situated within the drainage basin of the streams Dengora and Keta which join to
form the Sakelo, which in turn flows into the Akaki river. The surface catchment area is bounded eastwards
and southwards by a topographic limit: volcanoes Mt. Furi, Mt. Yerer, Gara Bushu, Mt. Guji and Mt.
Bilbilo. Geomorphology of the area is shown in figure 15, which is done by constructing shaded relief map
based on SRTM DEM.

Figure 14. Drainage and urban development of Addis Ababa area

2.5. Hydrochemistry
Water chemistry data can be used to infer groundwater flow directions, to define intended use of water,
identify sources and amount of recharge and to define local and regional flow systems (Anderson and
Woessner, 1992). Water quality is the constituents of water as affected by natural processes and human
activities and determines the intended use of water (Strickland et al., 1997). The major inorganic
constituents of water originate when water in the form of precipitation dissolves atmospheric gases such as
carbon dioxide and reacts with minerals on the surface of the earth (Freeze and Cherry, 1979). The quality
of groundwater first depends on the composition of the recharge water, the interactions between the water
and the soil, soil-gas and rocks with which it comes into contact in the unsaturated zone, and the residence
time and reactions that take place within the aquifer (Freeze and Cherry, 1979).

20
 Figure 15. Geomorphology of the Akaki catchment

Figure 16. Landsat imagery with shaded relief map

2.5.1. Water quality parameters

The water quality parameters considered during assessment of water quality include nitrate, chloride, pH,
acidity, alkalinity and electrical conductivity as well as major cations and anions. According to (WHO,
1993), the water quality parameters concentration range aims at safe guiding the health of human being as
well as other water use like irrigation, industry and domestic use.

2.5.2. Field sampling procedure

Water samples were taken from both boreholes and river waters. Certain parameters such as temperature,
pH, electrical conductivity, etc are difficult to preserve during the storage and are measured in situ in the

21
field. Duplicate samples were taken from each point for major ions determination in the laboratory by
adding acids in the field for preservation purpose. Sample containers, which do not interact with the
constituents intended to be determined during storage, were chosen. It is common practice to use filtered
(0.45 µm membrane filter) samples for metals in order to ensure that only dissolved species (which take
part in most geochemical reactions and are used in chemical equilibrium equations) and not suspended
constituents, which may be contributed from the wells are determined. So samples were filtered by this way
and almost all suspended clays and other materials are removed.

2.5.3. Chemical analysis results

To check the quality of measurements done in the laboratory there are various methods used to indicate the
correctness of the results (See figures 15, 16 and 17). Table 4 shows the quality check results according to
Hounslow (1995).

2.5.4. Reliability check

The accuracy of the water analysis was checked with the anion-cation balance. Water is naturally balanced
system. Ideally the sum of anions expressed, in meq/l must be exactly equal to the sum of cations, in meq/l,
in any sample. The principle of the anion–cation balance is that the sum of cations and sum of anions are
equal because the solution must be electrically neutral. Therefore, In a electrically neutral solution, the sum
of the cations should be equal to the sum of anions in meq/l which is also equal to EC/100 (μs/cm)
(Hounslow, 1995).

Electroneutrality (%) =
¦ Cations − ¦ Anions *100 (1)
¦ Cations + ¦ Anions
Based on the electroneutrality, analysis of water samples with a percent balance error <5% is regarded as
acceptable (Fetter, 2001). But in very dilute or saline water, up to 10% error may be considered as
acceptable due to the errors introduced in measuring major ions in dilute groundwater or in the multiple
dilution require for analysis of concentrated groundwater (Fetter, 2001). The cations-anions balance results
are found to be acceptable as the balance does not deviate from the 10% criterion (see table 4 below). The
analysis results indicate that the dominant dissolved cations in the groundwater of the area are Na+, Ca2+,
and Mg2+ with lower levels of K+. And the major dissolved anions in the groundwater include: HCO3-1.
This gives mixed cation bicarbonate water in the area. Table 3 and 4 below shows that in general
groundwater quality is fair.

Table 3. ITC laboratory chemical analysis result (2008)

Water points Ca (mg/l) Mg(mg/l) K(mg/l) Na (mg/l) Cl (mg/l) SO4 (mg/l) HCO3 (mg/l) NO3 (mg/l) PO4 (mg/l) EC (µs/cm)
EP-04 53.0 37.9 10.2 43.9 11.7 5.0 366.1 19.9 0.1 448
BH-05 77.1 35.4 8.7 35.4 4.9 5.5 402.7 24.4 0.2 530
BH-01 53.2 30.8 9.7 37.5 3.2 4.0 353.9 17.7 0.6 438
BH-10 55.2 35.3 8.4 29.6 5.0 4.0 366.1 18.6 0.7 458
F3 44.8 26.3 12.9 70.4 11.7 7.5 390.5 19.5 0.2 545
Iron & steel BH 92.3 39.7 10.6 47.7 9.2 10.5 518.7 19.5 0.3 646
Akaki metal products Factory BH 57.3 29.7 10.8 64.2 23.7 7.5 408.8 20.4 0.2 510
Spare Parts BH 73.2 30.6 7.4 44.1 6.6 7.5 396.6 18.0 0.3 470
Fanta Minch/ spring (Cs-1) 61.3 31.8 8.6 44.5 20.0 7.5 384.4 19.0 0.1 536
Dongora river 90.2 26.9 8.8 41.0 18.4 13.5 396.6 3.1 0.0 572
Fanta river 73.0 15.0 9.9 45.3 7.1 6.5 353.9 7.1 0.0 437
Fanta river near tsebel spring 54.3 11.6 8.5 34.3 4.7 5.5 305.1 6.2 0.0 442
Big Akaki, at theAkaki bridge 58.8 12.2 14.5 60.4 22.4 13.0 317.3 13.7 1.5 460
Kebena River at Bole Bridge 60.8 15.7 16.0 47.6 41.0 12.0 305.1 26.1 1.9 539
Rain water (RW-1) 5.6 0.6 5.3 4.2 0.7 0.0 36.6 0.0 0.0 104

22
Table 4. Ionic balance of the water samples
Water points Ca+2 Mg+2 K+1 Na+1 Cl-1 SO4-2 HCO3-1 NO3-1 Po4-3 EC/100 Ȉanion Ȉcation balance
(meq/l) (meq/l) (meq/l) (meq/l) (meq/l) (meq/l) (meq/l) (meq/l) (meq/l) (us/cm)
EP-04 2.6 3.1 0.3 1.9 0.3 0.1 6.0 0.3 0.0 6.5 6.8 7.9 -8.0
BH-05 3.8 2.9 0.2 1.5 0.1 0.1 6.6 0.4 0.0 7.7 7.3 8.5 -8.1
BH-01 2.7 2.5 0.2 1.6 0.1 0.1 5.8 0.3 0.0 6.4 6.3 7.1 -5.9
BH-10 2.8 2.9 0.2 1.3 0.1 0.1 6.0 0.3 0.0 6.6 6.5 7.2 -4.5
F3 2.2 2.2 0.3 3.1 0.3 0.2 6.4 0.3 0.0 7.9 7.2 7.8 -3.9
Iron & steel BH 4.6 3.3 0.3 2.1 0.3 0.2 8.5 0.3 0.0 9.4 9.3 10.2 -4.7
Akaki metal Factory BH 2.9 2.4 0.3 2.8 0.7 0.2 6.7 0.3 0.0 7.4 7.9 8.4 -3.1
Spare Parts BH 3.7 2.5 0.2 1.9 0.2 0.2 6.5 0.3 0.0 6.8 7.1 8.3 -7.3
Fanta Minch/ spring (Cs-1) 3.1 2.6 0.2 1.9 0.6 0.2 6.3 0.3 0.0 7.8 7.3 7.8 -3.3
Dongora river 4.5 2.2 0.2 1.8 0.5 0.3 6.5 0.1 0.0 8.3 7.4 8.7 -8.5
Fanta river 3.6 1.2 0.3 2.0 0.2 0.1 5.8 0.1 0.0 6.3 6.3 7.1 -6.4
Fanta river near tsebel spring 2.7 1.0 0.2 1.5 0.1 0.1 5.0 0.1 0.0 6.4 5.3 5.4 -0.2
Big Akaki at akaki bridge 2.9 1.0 0.4 2.6 0.6 0.3 5.2 0.2 0.0 6.7 6.4 6.9 -4.2
Kebena River at Bole Bridge 3.0 1.3 0.4 2.1 1.2 0.2 5.0 0.4 0.1 7.8 6.9 6.8 0.6
Rain water (RW-1) 0.3 0.0 0.1 0.2 0.0 0.0 0.6 0.0 0.0 1.5 0.6 0.6 -1.6

12
Sum of cations (meq/l)

10 y = 1.122x - 0.1183
2
8 R = 0.9655
6
4

0
0 2 4 6 8 10 12
Sum of anions (meq/l)
Figure 17. Graph of regression analysis of sum of anions against sum of cations

10
9 y = 1.0438x - 0.6322
2
8 R = 0.9318
EC/100 (us/cm)

7
6
5
4
3
2
1
0
0 2 4 6 8 10
Sum of anions (meq/l)
Figure 18. Graph of EC/100 (µs/cm) against sum of anions (meq/l)

23
 12
10 y = 1.1517x - 0.6939
2
EC/100 (us/cm)

R = 0.8701
8

6
4

2
0
0 2 4 6 8 10 12
Sum of cations (meq/l)
Figure 19. Graph of EC/100 (µs/cm) against sum of Cations (meq/l)

2.5.5. Presentation of results

Piper diagrams show the effect of various factors, including major ion composition of possible source
waters, as well as the proportions of mixing between those water sources in the samples. The effect of
geochemical interactions between water and aquifer materials also can be understood from Piper diagram.
The analysis result of the major anions and cations are plotted in Piper diagrams and Stiff diagrams for
determination of the water type. As discussed in Fetter (2001), Stiff diagrams are useful in making a rapid
visual comparison between water from different sources. The water samples from the area are presented in
the Piper. Selected representative samples were also plotted on Stiff diagrams (See Fig 20 and 21 for
details).

2.5.6. Water type deduction

The chemical laboratory results for all sampled points were imported in AQUACHEM v.5.1 software.
Aquachem is a software package developed specifically for graphical and numerical analysis and of water
quality data. The Piper diagram for 2008 dataset is shown in Figure 19. Groundwater in the basin is highly
mineralized with mixed cations with dominant cations being calcium, sodium and magnesium respectively
and bicarbonate as anion. The water type deduced from the Piper plot is mixed cation bicarbonate water
type, typical of shallow fresh groundwater.
Table 5. Water type of the present study in the catchment (2008)
Water point X Y Water type
EP-04 479118.5 977807.4 Mg-Ca-Na-HCO3
BH-05 476680.7 975819.8 Ca-Mg-HCO3
BH-01 478061.4 975076.1 Ca-Mg-Na-HCO3
BH-10 479155.1 976241.9 Mg-Ca-HCO3
F3 479779.2 981769.5 Na-Ca-Mg-HCO3
Iron & steel BH 476743.0 980936.4 Ca-Mg-Na-HCO3
Akaki metal products factory BH 476760.5 980804.8 Ca-Na-Mg-HCO3
Spare Parts BH 478347.8 977935.0 Ca-Mg-Na-HCO3
Fanta minch/ spring (Cs-1) 481354.8 981983.2 Ca-Mg-Na-HCO3
Dongora river 480153.1 977540.5 Ca-Mg-Na-HCO3
Fanta river 480856.6 981603.1 Ca-Na-HCO3
Fanta river near tsebel spring 478892.9 981633.9 Ca-Na-HCO3
Big Akaki at theAkaki bridge 476505.7 981305.8 Mg-NO3-Cl
Kebena river at Bole bridge 475606.3 994069.1 Ca-Na-HCO3
Rain water (RW-1) 472708.4 996421.5 Ca-Na-K-HCO3

24
The Piper diagrams are widely used for interpretation and classification of water. The concentration in
meq/l of major cations and anions from different water points analyzed in the area are plotted in the Piper
diagram. From the diagram two types of water can be identified:
1) Mg-NO3-Cl water type
2) Mixed cation bicarbonate water type
Legend
Legend
$ $ "Big Akaki river at akaki bridge
80 80
% Akaki metal products Factory BH
60 60
& BH-01

40 40 ' BH-05

( BH-10
20 20
) Dongora river

* EP-04
Mg SO4
) 0 + F3
'
(
/6
*,
&.-% , Fanta Minch/ spring (Cs-1)
80 80
+ - Fanta river

60 5 60 . Fanta river near tsebel spring

/ Iron & steel BH

40 (&* 40
'
/6, 0 Kebena River at Bole Bridge
) %+ 5
20 .0
- $ 20 rain water (RW-1)
$ 5 6 Spare Parts BH
/
6
.
-
'
&
(
5+)
*% 0
,
20

40

60

80
80

60

40

20

Ca Na+K HCO3 Cl
Figure 20. Piper Diagram showing the distribution of water types for water points

Akaki Bridge reveals a Mg-NO3-Cl and other rivers in the area are of Ca-Mg-Na-HCO3 and Ca-Na-HCO3
type. A shift of water chemistry towards SO4+ Cl and NO3 is mostly related to the infiltration of
contaminants into the rivers or subsurface rather than natural dissolution processes. The two Stiff diagrams
below in Fig. 21 indicate the two water types. All groundwater samples are more or less similar with the
Dongora river stiff diagram.
"Big Akaki Dongora river

Na Cl Na Cl

Ca HCO3 Ca HCO3

Mg SO4 Mg SO4

40 32 24 16 8 8 16 24 32 40 (meq/l) 40 32 24 16 8 8 16 24 32 40 (meq/l)

Figure 21. Stiff patterns of water samples from Akaki catchment

25
2.6. Subsurface conceptualization of the model domains
The area is made up of multi-aquifers having different hydrogeological characteristics. The existing data
shows that the area is composed of both inter-granular and fracture porosity type aquifers. The fractured
aquifers are composed of different volcanics such as weathered and fractured basalts, ignimbrites,
trachytes, rhyolites, and welded tuffs. Basalt and ignimbrites found in the upper Awash basin are mostly
tectonized and fractured; and therefore have good permeabilities (Berhane Melaku, 1982) as cited in
AAWSA(2000). All types of aquifers are found in the area: i.e confined, semi-confined and leaky confined
aquifers, including perched and leaky aquifers (AWSA-Seureca, 1991)

The depositions of alluvial materials between basaltic materials result in a large and very complex
geological structure with highly variable lithology. As a result the water level surface in the wells is not
everywhere smooth and continuous, and the discontinuities can be attributed to either topography or both
vertical and horizontal change in lithology. Perched aquifer conditions are recognized by the position of the
water strike and the recorded static water level. Static water level is below the water strike position in semi-
confined conditions.

In the research area, the geological age of volcanic rocks varies from Young Volcanics (Pliocene) to Alaji
Series (Lower Miocene) and the age of the rocks is becoming younger towards the rift system. Moreover,
the elevation of the area decreases to the rift as well. These variations of age and elevation affect the
weathering and fracturing of volcanic rocks to give variable recharge and aquifer characteristics. Moreover,
the structures are also affected by the distribution of groundwater and its location in the area (e.g. Filwoha
Fault). All the above factors affect the groundwater distribution and aquifer characteristics of the area.
Therefore, defining aquifer geometry is a very difficult task that remains ahead.

Groundwater resource evaluation and requires knowledge of various physical properties and hydraulic
parameters of aquifers and aquitards which includes hydraulic conductivity, storativity, specific yield, etc.
In order to define these parameters, in addition to pumping tests of various types, sampling and water
analysis in individual layers of water bearing formations is crucial. All the wells in the Akaki wellfield are
concentrated within small areas without having sound scientific justification. In most cases, therefore, a
hydrogeological description and interpretation is made by qualitative analysis of the surface information.
Figure 22 shows the North-South well log correlation in the Akaki catchment after Dereje Nigusse, (2003).
The figure shows the irregular layers of clay, porous rock, gravel, weathered regolith and hard rock. The
highly variable static water levels are also shown. Figure 23 below shows a detailed map of the Akaki
Wellfield area. Boreholes EP4 to EP8 are Akaki town water supply wells. The wells BH01 to BH26 are
Addis Ababa water supply wells. In this well group there is no BH15. Borehole BH 03b and BH 05a were
abandoned as they are found to be low yield and dry borehole respectively. Instead of these two boreholes;
BH03a, BH05b have been drilled. Four observation wells, MW01 to MW04 are monitoring wells found at
the vicinity of the Akaki River.

2.6.1. Well log correlations

According to Devis et al., (1966) as cited in Dereje Nigusse (2003), well log correlation helps in
determining the stratigraphic sequence of lava beds, buried soils and other features which extract influence
on the movement of groundwater. The lithologies of most geological formations tend to vary significantly
both laterally and vertically. The lithology, stratigraphy and the geological structures that affect the rocks

26
largely controls the nature and distribution of aquifers and aquitards in a geologic system. To investigate
these factors, borehole log is mandatory. However it is very difficult to get well documented and consistent
data for most of the boreholes that could provide full information about the subsurface lithostratigraphy
and characteristics of aquifers. See East–west well log correlation of Akaki catchment in Appendix 4. A
lithostratigraphic classification of the wellfield has been prepared using the information obtained from
more than 20 boreholes. The depth of production wells ranges from 119 m to 170 m and test wells from 60
m to 300 m from normal ground level.

Figure 22. North-South cross-section of well logs in the Akaki catchment (Dereje Nigusse, 2003)

27
Figure 23. Location map of the Akaki wellfield showing productive and monitoring wells.

The map also shows the location of cross-sections along N-S, E-W, NE-SW, NW-SE in the Akaki wellfield
(see also NW-SE and W-E cross-sections in Figures 23). N-S and NE – SW cross-sections are shown at
Appendix 3.

The diameter of wells drilled for water supply of Akaki and Addis Ababa is 24 inch from the ground
surface to a depth of 10 to 25 m, 17.5 inch in the middle part and 12.5 inch from around 80 to 100 m till the
bottom of the borehole. The observation wells have diameters of 12.25 inch from the ground surface to a
depth of 6 or 10 m whereas in the lower part the diameters vary from 9.6 to 8.5 inch.

The lithologic descriptions supported by resistivity and self potential log of boreholes along with the field
hydrogeological information have been used as a basis for classifying the hydrostratigraphic units. More
emphasis is given to the resistivity and self potential log than the strata description in the geologic log. Yet
the resistivity and self potential logging are only possible below the depth of static water level (See Figure
24 as an example for BH23). Self potential logging is used for acquisition of information concerning the
salinity of water in porous rock. As it can be proved from the geologic logs, the formation is basaltic rock
with fracture permeability. In water filled drill holes in rock with fracture permeability usually no
interpretable SP curve is attained (Repsold, 1989). In addition to this, results of water quality analysis show
fresh water. Therefore, the shape of the resitivity log curves depends mainly on the degree of fracturing and

28
the presence or absence of water. The water in the fissures contains ions such as Na+, Ca2+, Cl-, SO42- which
reduce the resistivity of the rock. Because of the limiting factor to use the spontaneous potential (SP) log in
this formation, resistivity log curves are mainly used in conjunction with the geologic log to determine the
stratification of the aquifers, aquitards and aquicludes in the wellfield. In some boreholes SP curves show
good correlation with resistivity curves.

In hydrogeological mapping and ground water modelling, aquifers, aquitards and aquicludes are defined
using the concept of hydrostratigraphic units by suppressing small local lithological variations. Several
geological formations may be combined into a single hydrostratigraphic unit depending on the scale of the
problem domain and the hydrogeological character of the rocks. Furthermore, water bearing zones often
depend on the degree of fracturing and to a lesser extent on lithology.

Based on the analysis of existing data as well as those gathered for this research, semi-confined (e.g. most
boreholes) and perched (e.g. almost the upper part of boreholes BH12 and BH13) aquifers within the
wellfield and confined aquifer close to Akaki bridge (located at x=476790, y=981229, z=2038m) have been
identified. The wellfield is characterized by multi-layered aquifers where the confining layers are
considered to be massive basalt and paleosols. The depositions of alluvial materials between volcanic
materials result in large and complex hydrogeological conditions with variable and unpredictable
lithostratigraphical correlation. The following figures of cross-sections show how the hydrostratigraphy of
the wellfield looks like by constructing diagrams from boreholes along NW-SE and E - W sections of the
wellfield (see Fig. 23 for the location of the cross-sections). For More hydrostratigraphic crossections and
well logs of the wellfield, see Appendix 1 and 3.

Based on the chemical and lithological variations, the aquifers are divided in to two broad classes; the
upper aquifer (<150m depth) and the deeper aquifers (up to 300m) (Demilie, 2007). The thickness of the
aquifer is not known since the drilling was terminated within the water bearing layers of the wellfield. The
types of aquifers are identified to be semi-confined and perched (Demilie, 2007).

29
 :$7(5,,,%25(+2/(%+
:(//,'(17,),&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB7<3(2)5,*BBBBBBBBBBB
63&+

$.$.,:(//),(/'$5($'
/2&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB67$7,&:$7(5/(9(/BBBBBBBB
P

$$:6$ ; 
2:1(5BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB:(//&225',1$7(BBBBBBBBBB
&203/$17
&2175$&725BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
< 

%+BBBBB
'$7('5,//('BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
(& = P
6&5((1'(37+ 0
6&5((16,=( LQFK
'5,//,1*6,=(
&$6,1*6,=(
'(37+P

'(37+P

(/(&75,&
LQFK

352),/( /,7+2/2*<
LQFK

/2**,1*&859(


 ([SDQVLYHVRLO




 :HDWKHUHGURFN

9 9 0DVVLYH%DVDOW

9 9 :HDWKHUHGURFN


9
 9
0DVVLYH%DVDOW
9


 :HDWKHUHGURFN

 $SSDUHQW5HVLVWLYW\
9
9 0DVVLYH%DVDOW      
9 P
 



 :HDWKHUHGURFN
)LUV W$TXLIHU






9 9 9
0DVVLYH%DVDOW

 9 9 9

:HDWKHUHGURFN
6HFRQG $TXLIHU
 
9 9 9
0DVVLYH%DVDOW

 :HDWKHUHGURFN
 7KLUG $TXLIHU

9 9 9 0DVVLYH%DVDOW





 :HDWKHUHGURFN
)RXUWK $TXLIHU





9 9 9
 
0DVVLYH%DVDOW
9 9 9








Figure 24. Geological and geophysical log of Borehole 23 (BH 23) in Akaki wellfield

30
Figure 25. Lithological cross – section along NW – SE in the Akaki wellfield (see Fig. 23)

Figure 26. Lithological X- sections along E-W in the wellfield (2.2 km) (see Fig. 23)

31
2.6.2. Well test analysis
In the Akaki wellfield, standard well tests were not carried out during water well drilling. In some cases
data recording have been stopped after a short period of time. In addition to this, partial penetration of
wells and uncertain boundary conditions are the main limitations for the right interpretation of diagnostic
plots. For example, Aynalem (1999) has identified various types of boundary conditions. These are
impermeable boundary in EP4, permeable boundary in BH18, casing and well storage effect in BH12.

For illustration purpose, some semi-log and log-log plots of a few boreholes has been presented below
(more pumping test curves are attached in Appendix 7). BH20 shows the leaky aquifer characteristics while
BH17 shows behaviour of unconfined aquifers. EP4 deviates from the diagnostic curves, perhaps because
of boundaries. The Akaki wellfield aquifer is mainly characterized as leaky-confined to semi-confined.

Figure 27. Semi-log plot of time-drawdown curve for selected wells

In the Akaki wellfield, boreholes to the north of Debrezeit road are reported by AAWSA (2000) to be dry
and unproductive. This could be because of the deep groundwater level along the Debrezeit highway, since
all the units above the static water level are composed of impermeable materials. More observations on the
situation here are given by the drilling contractor called COMPLANT during 2001.

2.6.3. Water level and abstraction monitoring

The general trend of the water level evolution in the wellfield from May, 1999 to January, 2002 is
presented below in Figure 29. The eleven wells designed for phase I, had been planned to begin pumping
between February and June 2001; however, due to technical problems the wellfield was not functional until
30th November 2001. Between 13 and 26 November 2001 trial pumping effectively started on the 30th
November 2001 and continued until 24th January 2002 when all the pumps failed and become defective
due to technical faults. Because of technical problems only eight out of eleven wells were operational
during the commissioning of the wellfield. These pumps have been removed from the wells for
32
replacement and in the mean time four smaller pumps have been air lifted and started pumping from 18th
March 2002. Therefore, the pumping at the wellfield can be roughly divided into four periods:
• Low abstraction (Jan 1999-Nov 2001) (5000 m3/day)
• Run trials (13 to 26 November 2001)
• Test by AAWSA (30th November 2001 to 24th January 2002) (25000 m3/day)
• Exploitation by AAWSA, which started from 18th March 2002 (>40000 m3/day)

Table 6. Summary of Phase I abstraction at Akaki Wellfield. (Nov 2001-Apr 2002)

Volume Average Contribution out
pumped discharge rate of the total
(m3) (l/s) pumped (%)
BH12 328943 26.0 16.3
BH09 491882 39.0 24.4
BH16 436396 35.0 21.6
BH17 410221 33.0 20.3
BH14 82434 6.5 4.0
BH08 227216 18.0 11.3
BH22 28287 2.2 1.4
BH18 12053 1.0 0.6

Total 2017435 160 100

Piezometers located around the wellfield were monitored to measure the piezometric impact (drawdown) of
the wellfield in operation. The data collected are used to identify directly hydrodynamic parameters of the
aquifer near the wellfield and above all to calibrate more accurately the model. More than 4000 water
levels have been collected since June 1999 to end of April 2002 (including the monitoring data obtained
from AAWSA between May 1999 and June 2001). Since the monitoring has started before the start of the
pumping operation, the data shows clear picture of the groundwater level evolution. When the wellfield
was pumped, around 20 piezometers were measured weekly and daily.

1100000
1000000
Production (m3)

900000
800000
700000
600000
500000
400000
Jul-02

Jul-03

Jul-04

Jul-05
Jan-03

Apr-03

Jan-04

Apr-04

Jan-05

Apr-05
Oct-02

Oct-03

Oct-04

Oct-05

Month
Figure 28. Akaki wellfield actual monthly production from July 2002 – October 2005

33
 Table 7. Annual production figures of the Akaki Wellfield (2002-2005)

Yearly
No Year Production Monthly average Daily average
3 3
(m ) (m ) (m3)
1 2002 3210296 267525 8917
2 2003 8368264 697355 23245
3 2004 11054472 921206 30707
4 2005 11323801 943650 31455
5 2006 0 0
6 2007 0 0

Evolution of the piezometric levels [01 June 1999 - 31 January 2002]

2020.00
BH01b BH04 BH05b BH06 BH07 BH08 BH09 BH10 BH20 MW01b

2019.75
Heads(masl)

2019.50

2019.25

2019.00

2018.75

2018.50
6/1/1999 8/31/1999 11/30/1999 2/29/2000 5/30/2000 8/30/2000 11/29/2000 2/28/2001 5/30/2001 8/29/2001 11/29/2001
Time

Figure 29. Evolution of groundwater levels from May 01, 1999 to January 24, 2002

The response of the water level in the wellfield during the rains and after the rainy season is not significant
although some small peaks appear between July and September (Figure 29). The water level shows
continuous decline with little peaks in the months between July and October. In general over the last 3
years the water level in the wellfield declined for an average of about 0.15m. The change in the
groundwater level during the pumping period for 56 days (beginning of 2002) is shown in Figure 30. The
groundwater level declined with higher slope until the 24th of January corresponding to the high rate of
pumping (average of 260 l/s). After stoppage of pumping for 53 days the level tends to recover to its
natural value, however the recovery rate was slow and it did not recover to its original level before
beginning of pumping. The water level again started to decline after March 18th (beginning of the second
pumping period) with a smaller slope as compared to the first pumping period because of the smaller
abstraction rate about 193 l/s.

The situation became more steady after the well testing in the beginning of 2002. As can be seen from the
water level data logger shown in Fig. 30, continuous decline of water level for one meter is identified
between the periods of August, 2002 to Sept, 2004. From September, 2005 to November, 2005 The
34
decrease in water level is in the order of 2 m/year or slightly less due to 950000 m3/month.The continuous
decline without much fluctuation seems to indicate that seasonal recharge effects are not important in the
wellfield area or abstraction is much larger than recharge.

2019.75
BH01b BH04 BH05b BH06 BH07

BH08 BH09 BH10 BH20 MW01b

2019.50
heads (masl)

2019.25

2019.00

2018.75

2018.50
21/11/01

11/12/01

31/12/01

9/2/02

1/3/02
1/11/01

20/1/02

21/3/02

10/4/02

30/4/02
Time
Figure 30. Evolution of the piezometric levels during the pumping (1/11/2001-30/ 04/2002)

2018
MW-01
MW-02
2017 MW-03
MW-04
2016 TW-01
heads(masl).

2015

2014

2013

2012
2002 2003 2004 2005 2006
Time (years)

Figure 31. Akaki wellfield water level monitoring report, 2002 to 2006, (see fig.23).

35
Figure 32. Hydrographs of MW1 (top) & TW1 (2005 - 2007 (bottom)

36
3. Groundwater flow modelling of Akaki
catchment

3.1. Introduction
Groundwater modelling is used to make predictions about a groundwater system’s response to a stress, to
further increase our understanding on the hydrological system in the catchment and its wellfield. The
groundwater model was used as a tool for understanding the system, its behaviour and for predicting its
response to external stresses. The developed model serves as a tool to improve our understanding on the
catchment’s groundwater flow system. In this study, steady state models are developed for the aquifers in
the Akaki catchment, and the Akaki wellfield in particular. The results of the wellfield model will then be
used in contaminant transport analysis.

3.2. Development of conceptual model

Completion of model conceptualization process is necessary prior to determining the modelling approach
to be used. After building the data base required for the model, the next step is developing a conceptual
model. This is the most important part of the modelling process. The key data requirements in the process
of conceptualization are selecting data on hydro-stratigraphic units, surface water bodies, recharge and
discharge zones. The use of model conceptualization is to create a simplified model of the system to be
simulated, by simplifying a system to an extent that a logical model approach with appropriate model
algorithms can be defined (Rientjes, 2007). Conceptual model of the catchment in figure 33 describes how
water enters an aquifer system, how it flows through the aquifer system and how it leaves the aquifer
system.

Figure 33. N–S conceptual model of GW flow in the Akaki catchment (Demlie, 2007)

37
 Table 8. . Parameters considered in conceptualization of the system (Aquaterra consultant, 2000)

3.3. General Assumptions

Several general simplifying assumptions are introduced to simulate groundwater flow, and to develop and
to calibrate the model. These include:
• Fractures and weathered zones through which water flows are considered as porous medium to which
Darcy’s Law (Getachew Asmare, 2005) can be applied.
• Net recharge from precipitation is not spatially uniform because there is heterogeneity in the spatial
distribution of hydraulic conductivity, geology, total precipitation, and slope;
• Aquifer heterogeneity, vertical anisotropy, and the presence of fracture and faults impact the spatial
distribution of hydraulic conductivity. Consequently, a zonation approach is adopted where similar
hydraulic conductivity values are assigned to specific regions on the basis of above factors.
• Vertical flow is assumed to be negligible since one aquifer system/single layer is considered.
Different boundary conditions are assumed in the model: head dependent boundary is used on the wet
land of Akaki and Akaki river itself and flux boundary is assumed to the north, NE, and NW. Drain
package is assigned to small reservoirs.

3.4. Governing Flow Equations

The three-dimensional finite-difference groundwater flow model of Harbaugh et al., (1998) was used for
the computer simulations. The mathematical model simulates flow indirectly by means of a governing
equation thought to represent the physical processes that occur in the system. This model is based on the
following governing equations for anisotropic, heterogeneous aquifer. Steady state and transient flow
equations are shown in (2) and (3) respectively.
∂ § ∂h · ∂ § ∂h · ∂ § ∂h ·
¨kx ¸ + ¨ky ¸ + ¨ kz ¸ − W = 0 (2)
∂x © ∂x ¹ ∂y ¨© ∂y ¸¹ ∂z © ∂z ¹

38
 ∂ § ∂h · ∂ § ∂h · ∂ § ∂h · ∂h
¨kx ¸ + ¨¨ k y ¸¸ + ¨ k z ¸ −W = Ss (3)
∂x © ∂x ¹ ∂y © ∂y ¹ ∂z © ∂z ¹ ∂t

Where
H = Potentiometeric head (L);
W = Volumetric flux per unit volume and represents sources and/or sinks of water (T-1);
Ss = Specific storage of the porous material (L-1);
T = Time (T).
kx, ky, and kz are values of hydraulic conductivity along the x, y, and z coordinate axis, which are assumed
to be parallel to the major, axes of hydraulic conductivity (LT-1);

In general, Ss, kx, ky, and kz are functions of space (Ss=Ss (x, y, z), kx=kx (x, y, z), etc) and W is a function
of space and time (w = w (x, y, z, t). The set of algebraic equations that result when approximating a
groundwater flow model using the method of finite differences is normally solved using a combination of
matrix and iterative solution techniques (Anderson and Woessner, 1992).

3.5. Modelling Approach

The Processing Modflow (PM) software (Version pm5), developed by the United States Geological Survey
and Chiang and Kinzelbach (1998) has been used for the construction of the flow model of the study area.
A three-dimensional model grid was used to represent a two-dimensional aerial flow through a single layer.
This numerical groundwater modelling software which is an enhanced version of Processing Modflow is
developed by Web Tech 360 Inc. in 2002-2003 based on the original work of Chiang and Kinzelbach
(1998).
In order to use a finite difference approximation, a grid is superimposed over the digital elevation model of
the study area, and aquifer hydraulic parameters necessary to solve the flow equation are averaged over the
area of cell or grid block and assigned at a node at the center of the block. The finite-difference method is
used to compute the average head value in all model cells. In the block-centered formulation, the nodes for
which water levels are simulated are located at the center of the grid cells. These cells are the smallest
volumetric units over which the hydraulic properties are assumed constant.
PMWIN requires the use of consistent units throughout the modelling process. For instance, if one use
length (L) units of meters and time (T) units of seconds, hydraulic conductivity will be expressed in units
of (ms-1), and pumping rates will be in units of (m3s-1).

3.6. Geographic Extension of the model

The model area encompasses the limits of the regional flow system of the Akaki river and to the south
extended up to the Abba Samuel gorge. The northern, western and eastern catchment boundaries of the
Akaki River are taken as no-flow boundary. General head boundaries are used at the outlet of the
catchment below Abba Samuel lake at the gorge where springs are emerging to simulate the amount of
water flowing out of the catchment.

3.7. Spatial discretization of Model Grid

The catchment boundary was delineated by the application of DEM hydro processing package in ILWIS
software. The extraction operation constructs catchments: these are calculated for each stream found in the
output map of the drainage network ordering operation (ITC, 2001).

39
In MODFLOW, an aquifer system is replaced by a discretized domain consisting of an array of nodes at
which hydraulic heads are calculated and associated in finite difference blocks, cells (Lubczynski, 1997,
Rientjes, 2007).

The model area encompasses the limits of the flow system of the Akaki river and extends up to Mt. Bilbilo
and Guji to the south. The model spans an area of 1500 km². The model grid consists of 150 columns and
135 rows with 20250 active cells and regular grid spacing of 400 m x 400 m. The geographic boundaries
of the model grid were determined by using ASTER DEM as background map. The image map was
projected in metric coordinates (UTM) and Adindan datum, and then imported into MODFLOW. A finite-
difference grid superimposed over a 1500 km2 area was designed and constructed based on the
simplification of a conceptual model representing the physical properties of the groundwater system.
Fig. 34 represents this three dimensional conceptual geospatial model frame work showing the number of
columns, rows and layers used in the current model.

3.8. Model structure

The aquifers within the model area are very complex and cannot be separated and it is assumed that they
are hydraulically connected. The geological logs of the boreholes and cross-sections in the model area and
the wellfield show high irregularity of different lithological units. They do not have sufficient extent to
justify a multilayer representation of the reservoir. Also due to the fact that the study is for long term
pumping, it is assumed that multilayer effect will have very little consequence on the outcome of the
model. Therefore, a single layer structure is adopted for the model. The horizontal variation can be
accommodated by the variable hydraulic properties. Although there is vertical variability it is not
considered in the model.

Figure 34. 3D conceptual geospatial model frame work of Akaki catchment

The aquifer in the lower basalt is assumed to be semi-confined. The alluvial sediment is a local unconfined
aquifer with intergranular porosity. In order to identify the groundwater flow it is necessary to identify

40
hydro-stratigraphic units (number of aquifer, aquitards, aquicluide, etc.) that have common physical and
chemical characteristics.

The geometry of the aquifer systems in the catchment area is highly variable, and discontinuous.
Nevertheless, the model was developed with a single layer for the following reasons:
ƒ The aquifer in the area, particularly the scoria and scoraceous basaltic lava flows are not a single
continuous unit. They consist of small discontinuous patches representing local flows. Therefore
their effect disappears on a larger scale.
ƒ The water level of the aquifers deeper than 170 m rises and maintains the regional water level
elevation around Akaki Wellfield. This indicates that the water in the aquifers deeper than 170 m is
hydraulically connected with the water of the shallow aquifer. Therefore, almost all of the aquifer
parameters are not defined for individual layers. They are obtained as a cumulative effect or total
effect of all layers and at this stage it is not practical to treat them separately as individual layers.
ƒ The available data show that the geological conditions are very complex and assumption of
different model layers is impossible because of
The non uniform and highly variable and complex nature of volcanic aquifer related to lava
flows
The fracturing of the rocks
The scarcity of adequate well test data from subsurface geological layers separately
ƒ Vertical flow is assumed to be negligible since one aquifer system/single layer is considered.

3.9. Physical parameters of the system

The model should generally have values specified initially for all necessary parameters. The spatial input
variables are initial hydraulic head, horizontal anisotropy, horizontal and vertical hydraulic conductivities,
recharge, discharges, and effective porosity. Therefore, values were assigned to each active grid cell, based
on its location within the study area.

3.9.1. Initial hydraulic heads

Initial conditions refer to the hydraulic head distribution in the system at the beginning of the simulation
(Anderson and Woessner, 1992). For the present case, the static water level records of the wells are
interpolated within the model to obtain the initial hydraulic heads for the entire model. Initial hydraulic
heads distribution in the study area was obtained from the well database in the area.
The hydraulic heads distribution was interpolated first using ILWIS and imported to MODFLOW at the
beginning of the model run and each cell in the model was assigned its head value. This initial data can
also be interpolated using kriging method (Harbaugh et al., 2000) in MODFLOW.

The ground elevations at each model cell are extracted from corrected ASTER DEM of the area, and then
depth-to-static-water level values from the well database is subtracted from each DEMs obtained. The
results of this calculation are used as initial prescribed hydraulic heads in each cell for the initial
specification of head values. Similarly, other input maps are prepared by using ILWIS software and are
loaded to MODFLOW in the form of matrix to be used in modelling process.

41
 Figure 35. Initial hydraulic head values of the catchment

3.9.2. Hydraulic conductivity

Hydraulic conductivity refers to the ability of the aquifer materials to transmit water, which in turn controls
the rate at which groundwater will flow under a given hydraulic gradient. Hydraulic conductivity is
important because it controls the rate of groundwater movement in the saturated zone, thereby controlling
the degree and fate of contaminants.

3.9.2.1. Horizontal Hydraulic Conductivity

One of the important hydraulic parameters required for a steady-state groundwater flow model is either
transmissivity or hydraulic conductivity in a distributed fashion cell by cell across the model grid. Zonation
for the input parameters was carried out based on geological information, point hydraulic conductivity and
transmissivity data of the pumping tests. Initially, the hydraulic parameters estimated from pumping test
results of previous studies were applied; later the parameters were adjusted during the calibration process.

To specify the horizontal hydraulic conductivity, initially the hydraulic conductivity map was produced
from hydraulic conductivity values of boreholes distributed in the area. The hydraulic conductivity data
were assigned by interpolating the pumping test data in ILWIS and classifying certain zones based on the
geology, hydrogeologic and geomorphologic conditions (Fig.36).

42
 Figure 36. Hydraulic conductivity values of the catchment

Then, the map is overlaid on the model grid, and then obtained respective average values are assigned to
each model cells. Higher hydraulic conductivity values are obtained in the wellfield (26 m/day), Smaller
values are obtained north, northeast, and west of the wellfield with the minimum value reaching 1 m/day
(see Fig.36).

3.9.2.2. Vertical Hydraulic Conductivity or Leakance

PMWIN Pro calculates vertical leakance by using the following rules. Vertical leakance between the layers
i and i+1 is given as the value of the ith layer. Therefore, since MODFLOW assumes that the bottom layer
is underlain by impermeable material; the leakance data are not required for the bottom layer of multiple
layer models or for a single layer model like the one currently dealt with. Because of this, vertical hydraulic
conductivity is not required for this single layer models.

3.9.3. Horizontal Anisotropy

Horizontal anisotropy is the ratio of the horizontal hydraulic conductivity along columns to horizontal
hydraulic conductivity along rows i.e., the ratio of (Kh) along columns to (Kh) along rows. They are
assumed to be equal to 1 representing that both hydraulic conductivities along Eastings and Northings at
particular location or cell are the same. As the result of this, no anisotropy was assumed in the models.

3.9.4. Aquifer geometry

The demarcation of the aquifer extent was made based on the available technical details of the boreholes.
Top of aquifer elevation was assigned based on the static water level of the area. The bottom elevation of
the aquifer was obtained by subtracting one hundred meters from the top of aquifer.

43
 Figure 37. Bottom of the aquifer

Figure 38. Top of the aquifer

3.9.5. Aquifer thickness

Aquifer thickness of the catchment is obtained by subtracting the top of aquifer from bottom of aquifer.
Aquifer is found to have an average thickness of 100m.

44
 Figure 39. Vadose zone thickness map

3.9.6. Transmissivity
The transmissivity (T-value) calculated from pumping tests in the Akaki catchment and wellfield reveal
that it ranges between 6 m2/day to 105408 m2/day. To the north of Akaki wellfield and Kality area, T-
values range from 3 - 6m2/day (Yirga Tadesse, 2004 as cited in Gizaw, 2002).

The transmissivity values are converted to hydraulic conductivity by dividing T-values with aquifer
thickness. The calculated hydraulic conductivity values vary between 7 to 675 m/day in the wellfield.
These values lie in the range of permeable basalt (Aynelam, 1999). Low transmissivity value is obtained
from the EP wells located at the north-eastern part of the well filed. Generally the aquifer shows a gradual
increase in transmissivity towards the wellfield. Therefore, the wellfield has relatively high value but it is
surrounded by rocks of low values showing that the highly productive aquifer at the wellfield has very
limited extent (Fig.39). Transmissivity values were calculated by Logan’s approximation method.
according to Logan, 1964, from specific capacity by the following relationship.

T = 1.22 * Sc (4)

Where T = Transmissivity (m2/day) and SC = specific capacity (m2/day)

45
 120000

100000 T = 1.8819SC + 597.97

2
R = 0.9509
80000
T (m /day).

60000
2

40000
20000

0
0 10000 20000 30000 40000 50000 60000 70000
2
SC (m /day)
Figure 40. Comparison of Specific capacity and values in the catchment

100000
y = 0.6022x + 0.1
10000
T estimated (m /day).

2
R = 0.95
1000
2

100

10

0
0 1 10 100 1000 10000 100000
2
T calculated (m /day)
Figure 41. Comparison of pumping test and Logan estimated values in the catchment

Figure 42. Transmissivity map of the area.

46
3.9.7. Effective Porosity
Effective porosity is the sum of the interconnected pore space, i.e., excluding isolated pores. Total porosity,
on the other hand, is the volume of the reservoir rock which is fluid (water) filled, expressed as a
percentage or a fraction of the gross (bulk) rock volume. Although a steady state flow simulation model
does not require effective porosity parameter, it is necessary for the computation of travel times and
contaminant transport processes. However, determination of effective porosity requires core analysis
(humidity-dried or oven-dried) or log analysis (density log or neutron log) which is beyond the scope of
this work.
For the vast majority of rocks in the study area, effective porosity equals to total porosity, because most of
them are composed of non-clay minerals coarser than silt and at the same time they are highly affected by
tectonics. Though the total porosity of basalt is generally low, fractured and weathered basalt of Akaki
catchment area is assumed to have high effective porosity. The scoraceous deposits at Akaki wellfield are
assumed to have higher effective porosity values. The black cotton soil and scoria of the area constitute
higher proportion of medium sand and medium gravel respectively. According to Tenalem et al. (2005), a
representative value of the effective porosity for medium sand is 28% and medium gravel is 24%. The
mean value of the total porosity which is a measure of the existing voids, expressed as the ratio (in
percentage) of the volume of the voids, Vv, to the total volume, VT for medium gravel is 32 %, and for
medium sand is 39 %. Therefore, effective porosity of 30 % is used in the model.

3.9.8. Boreholes and observation wells

Most of the boreholes in the catchment are concentrated on the southern and central part. The data base of
the boreholes was collected from the concerned offices in Addis Ababa including AAWSA. There are no
proper data base management of the boreholes and it is difficult to obtain a continuous time series
groundwater level monitoring for almost all boreholes. In almost all wells, the pumping well serves as an
observation well except for six monitoring wells in the wellfield. In the area, the well abstractions were
implemented using the well package of MODFLOW.

3.9.9. Model Boundary Conditions

In modelling, we are interested in a specific part from the continuous real world system. Thus, the effect of
real world in terms of hydrological influences at the model boundaries must be described. Correct selection
of boundary conditions is a critical step in model design (Anderson and Woessner, 1992) , because the
boundaries largely determine the flow pattern. In other words, mass exchanges across the boundaries are
simulated and hence incorrect boundary conditions generate errors in the water balance of the system under
study. In groundwater flow system, we can use physical boundaries including impermeable geologic
formations and surface water bodies or hydraulic boundaries that include groundwater divides and flow
lines. For the case of Akaki catchment and wellfield, the boundary conditions will be selected based on the
actual ground field condition of the site.

For the present study specified flow (no flow) boundaries and head dependent flow boundaries (Cauchy
boundary conditions), river and drain packages were applied. A general head boundary package was
employed to simulate groundwater outflow through the southern outlet of the basin below Aba Samuel lake
for the catchment modelling. General head boundaries are normally used along the edge of the model to
allow groundwater to flow into or out of the model under a regional gradient. If the water elevation rises

47
above the specified head, water flows out of the aquifer. The expression applied for the head dependent
flow in the general head flow boundary is:

Qb = Cb × (hb − h ) (5)
Qb = Simulated hydraulic head (L)
Cb = Hydraulic conductance of the boundary (L2 T-1)
hb = Hydraulic head at or beyond the boundary, imposed head,( L)
h = hydraulic head in the aquifer (L)
As discussed in the conceptual model formulation, Akaki river is in hydraulic contact with the
groundwater. The flow of water between an aquifer and overlying river is commonly simulated using the
river package. (See Appendix 10). The expression for which river package is based on is:
Q RIV = C RIV × (H RIV − h ) h > RBOT (6)
Q RIV = C RIV × (H RIV − RBOT ) h ≤ RBOT (7)
C RIV = ( K × L × W ) / M (8)

Where,
QRIV = Rate of leakage between the river and the aquifer (L3 T-1),
CRIV = Hydraulic conductance of the river bed, (L2T-1),
HRIV = Head in the river (L),
H = Hydraulic head in cell (L),
RBOT = Elevation of the bottom of the riverbed (L),
K = Hydraulic conductivity of the riverbed material (LT-1),
L = Length of the river within a cell (L), W is the width of the river (L) and
M = Thickness of the riverbed (L). (See Appendix 9 for river package of the catchment)

The recharge to the aquifer from the loosing reaches of the river is assumed to take place by the vertical
areal recharge which is part of recharge mechanisms in the catchment.

To help visualize the model site, a DXF file or a raster graphics image of the area is used to overlay as a
site map with the locations of the boundaries and the pumping wells indicated. The base of the volcanic
aquifer is the basement volcanic rock where groundwater flow is negligible. At the bottom of the layer, the
no-flow boundary was assigned assuming that the boundary coincides with the fresh volcanic rocks. On the
left hand side of the model (western side), eastern, northern of the model area, no-flow boundary (Q = 0)
was assigned. It was assumed that the boundary of the model coincides with the water divides as they are
located along topographic high areas.

The river package is assigned to little and big Akaki rivers, which are the main rivers in the catchment.
Drain Package is assigned to small reservoirs such as Dire and Gefersa since they are feed by springs.
Constant head boundary is given to big reservoirs such as Abba Samuel and Legedadi lakes. These are cells
in which hydraulic head is kept fixed at a given value all over the entire simulation time.

48
Field hydrogeological observations, hydrochemical and isotopic studies indicate that groundwater flows
out of the southern tip of the basin (Gizaw, 2002; Demlie, 2007). This area is treated as a general head
boundary. General-head Boundary was given to outflow zone below Aba Samuel lake which is out let of
the catchment. The large reservoirs such as Aba Samuel and Legedadi are treated as constant head cells. A
total of 150 pumping wells are considered with a pumping rate varying between 10 and 7808 m3/day. There
are no recharging wells. By default and convention the area outside the model domain is deemed to be a
“no flow” and as such it is not necessary to set this area to inactive.

Table 9. Major types of model boundary conditions (Aquaterra consultant, 2000)

Boundary Type Technical Common Effects of Boundary Comment

Description Applications Condition on
Solution
Specified Head First Type Rivers, coastlines, Easiest to solve, but Commonly used
(the head value Or Dirichlet lakes, groundwater constrains solution because head data
is specified and Boundary divides, known to greatest degree can be measured
the model pumping water levels (can artificially much easier than
calculates in bores, dewatering constrain solution flow data. It
the flow across targets. too greatly). allows an
the boundary to inexhaustible
or from the amount of water
model domain) flow (calculated
by the model) into
or out of a model.
Specified Flow Second Type Impermeable Moderately The “no flow”
(the flow value or boundary, difficult to solve, boundary is a
is specified and Neumann groundwater divide or and involves special version of
the model Boundary streamline, infiltration moderate the specified flow
calculates source, evaporation constraints on boundary, used to
the head at the sink, lateral inflow or solution. define low
boundary) outflow, other known permeability
sink or source fluxes formations
(eg. adjacent aquifer adjacent to or
or pumping bore) underlying
aquifers, or for
streamlines
Head-dependent Third Type Leaky rivers, drains, Most difficult to Care is required in
Flow
Or Cauchy flow to or from solve, and involves some cases, as the
( model (mixed) adjacent aquifers, least constraints on model calculated
calculates the
Boundary basement leakage, solution. Can form flow is subject to a
flow for the
springs. a very complex and conductance
given head)
sensitive boundary parameter, and
condition. this may violate
some calibration
assumptions.

49
Figure 43. 3D view of the Akaki catchment

(m)

Figure 44. Boundary conditions of the model domain

50
3.9.10. Sinks and sources of the modelled area
Rainfall is the main source of water to the aquifer system of catchment. Recharge processes are explained
in section 3.9.11. Average recharge flux determined by CMB was compared to the other methods. CMB
results were compared with other model results and through application of recharge package in Modflow,
its average value was assigned to the model. The annual recharge value was converted into daily values.

Evapotranspiration and well abstractions are major sinks of the model for this particular catchment.
Total well abstractions of 23 MCM/yr, which may include private, factory and water supply wells of
AAWSA are applied to the model. The current abstractions may be closer to 50MCM/yr but there is no
well managed database of current abstraction. Springs and small reservoirs (Dire and Gefersa) are
simulated by giving drain package since they are feed by groundwater and used as starting point of rivers.
A head dependent boundary (outflow zone) was placed at Aba Samuel gorge spring sites south of wellfield.
Moreover, base flow to the rivers (effect of river heads to surface water) was simulated by river package in
the model (see groundwater water balance). Constant head was given to Aba Samuel and Legedadi
reservoirs. Groundwater evapotranspiration was ignored since it is assumed that there will not be much
groundwater loss from saturated zone in afro alpine climate at mountain top like Addis Ababa and deep
rooted trees like eucalyptus trees that are capable of abstracting large amount of groundwater are cut on the
past decades. Besides that there is also no data which is available on it in the area.

3.9.11. Recharge
Recharge is often the most important factor in evaluating regional aquifer systems in arid and semi arid
environments and it is unfortunately the most difficult to quantify (Wood and Sanford, 1995). Natural
recharge to the saturated zone in a groundwater reservoir comes from vertical percolation of precipitation
and from the losses of the streams flowing after important rains in the upper part of the catchment. Direct
recharge refers to precipitation that contributes to soil moisture content and crosses the water table and
becomes part of the groundwater flow system. The recharge input assigned was assumed to be spatially
variable. Recharge is assumed to take place in all areas. A positive value is used to represent recharge into
the aquifer system. Most recharge to the groundwater occurs only during the wet season. Since the rainfall
distribution in Addis Ababa is bimodal (Belg and Kiremt seasons), the annual total average recharge is
fairly distributed for these seasons with the other months having negligible recharge. Several methods for
estimating of groundwater recharge have been developed and can be divided into the physically based and
chemical and isotopic methods (Simmers, 1988).

Previous recharge studies in the catchment

Various attempts had been made to estimate groundwater recharge in the akaki river basin (AAWSA-
Seureca, 1991; Tahal 1992; AAWSA, 2000). As cited in Gizaw (2002), an earlier estmate (Tadese
Belachew, 1975) assumed that the net replenishment in the Aakaki basin is 95 mm, i.e. 7.3% of mean
annual rainfall of 1300mm. AAWSA(2000) estimated the mean annual recharge to be 105.4 mm using a
semi distributed water balance equation at annual time scale which is equivalent of 9.6 % of mean annual
rainfall in the catchment. Taking the area of the catchment (1464 km2) direct natural recharge volume is
154x106 m3. Demlie (2007) was found that the chloride mass balance estimate of recharge in the catchment
is 265 mm, which amounts to 24% of the weighted mean annual area precipitation of the catchment. He
suggested that the CMB method, with limited data, has over estimated the mean annual groundwater

51
recharge rate. Besides that Demlie (2007), did also groundwater recharge estimation using semi distributed
catchment water balance. He obtained total weighted mean annual recharge of 105.4 mm/yr which accounts
for 10% of the mean annual areal precipitation. AWSA (2000) found 51 mm/yr of recharge with a semi-
distributed water balance model as will be discussed later.

3.9.11.1. Groundwater recharge estimation by chloride mass balance method (CMB)

Groundwater resource studies require the estimation of the quantity of water moving downwards from the
soil zones as a potential recharge (Rushton et al., 2006). Yongxin and Beekman (2003) stated that the
chloride mass balance method was applied for recharge estimation worldwide in recent time. According to
Simmers et al. (1997), chloride is the most important environmental tracer and has been used to estimate
rates of groundwater recharge under a wide range of climatic, geologic and soil conditions. The most
widely used chemical technique to estimate groundwater recharge is the Allison and Hughes (1978)
developed chloride mass balance technique based on earlier work by (Eriksson, 1969 and Gieske, 1992).
This method is inexpensive, easy to use and gives reasonable results when reasonable data is available.
However, the method has its own limitations. It can not be used when other sources of chloride other than
only from meteoric water are present in the subsurface; such as from saline water or from dissolution of
evaporates (Beekman et al., 1997). The chloride mass balance method is applicable to various
environments (Gieske, 1992).
The method has several shortcomings, one of which is that the method cannot be used in environments
affected by other sources of chloride other than total atmospheric fallout. Thus the following assumptions
are made to apply the method.
ƒ Chloride is conservative and will not undergo any chemical reaction with the geologic material.
ƒ Precipitation is the only chloride source in groundwater
The basic equation used to calculate the annual groundwater recharge with the assumption negligible
chloride dry deposition in an area is based on equation 9.

Chloride in rainwater and groundwater

The chloride concentration in the rainwater and groundwater of the study area that are collected from the
catchment in Sept. 2008 are analysed in ITC, the Netherlands for the determination of recharge in
combination with the chlorides of 2008 determined by Demelie (2008).

Figure 45. Chloride map of groundwater in the Akaki catchment

52
3.9.11.2. Basic assumptions of chloride mass balance
The chloride mass balance is used to estimate average areal recharge to a groundwater basin. The
underlying assumptions for this method are:
ƒ The only source of chloride is rainfall
ƒ Chloride ion is not leached from the aquifer formation
ƒ The chloride ion behaves conservatively under steady state conditions and there are no sources and
sinks for chloride ion,
ƒ Recharge to the aquifer is only by direct precipitation on the surface and subsequently recharge
through unsaturated zone which assumes that rainfall is not carried out of the catchment by runoff,

Chloride meets these requirements since it is a conservative ion not taken up nor is it leached from
vegetation or aquifer formations in the study area in significant amounts. It is assumed that the cities
effluents, approximately 50 MCM/yr, are mainly drained by surface runoff. 10 % of the effluents is
assumed to be recharged back to aquifer (Tamiru, 2001). Up to today except 1000 tones per year from the
city resulting from urban return flow, no sources or sinks of chloride have been reported in the study area.
According to Edmunds et al. (1994) nutrient cycling of plants may interfere with the movement of the
solutes on the annual basis. However a steady state condition may be assumed in areas with stable
landscape, so that there is a balance between solutes removed by plant uptake and the amount of solutes
created by plant decomposition. This assumption only holds when there is no addition from fertilizers or
permanent removal by crop harvesting or transfer by grazing animals (Obakeng, 2000). When there is a
balance between the chloride flux at the surface and chloride flux below the evapotranspiration and mixing
zone, and the above discussed assumptions are met, the following mass balance excluding dry deposition
term was defined (Eriksson, 1969):

P × Cl p = RT × Cl gw (9)

Where P = Precipitation (mm/yr)

Clp = Chloride content in rainfall (mg/l)
RT = Total recharge rate (mm/yr)
Clgw = Chloride content in groundwater (mg/l)

Different flow components in the unsaturated zone contribute to the chloride concentration in the
groundwater. Total chloride deposition comprises both wet and dry deposition. Wet deposition refers to a
chloride content in rainfall while dry deposition is the chloride that falls with dust and aerosols during dry
season. Dry deposition can be estimated by rinsing the rain gauge or totalizer with deionised water during
dry periods (Gieske, 1992). Eriksson (1985) stated that the average groundwater chloride content is
calculated as harmonic mean of chloride concenterations. The formula is given by equation 10.

N
Cl gw = N (10)
1
¦
i =1 Cl gw

Where
Clgw = Groundwater chloride concentration of samples (mg l-1)

53
 N = The total number of observations
Based on the collected groundwater samples, the mean, maximum and minimum chloride content in the
groundwater of Akaki catchment is 6.9, 23.7 and 0.6 mg l-1 respectively.

Based on the method, the following results are obtained:

ƒ Average chloride concentration in rain water (0.8 mg l-1 )
ƒ Harmonic mean of chloride content in groundwater (3 mg l-1)
ƒ Average annual rainfall (1224 mm/yr)
ƒ The estimated recharge based on CMB is 326 mm year-1 which is 24 % of the average annual
rainfall, which in turn is 1224 mm/yr, in the area.
Thus the estimated recharge may range between 272 to 410 mm/yr depending on the range of chloride
concentrations in both rain and groundwater. This figure sets the upper limit to the recharge because
effluents from the city are also infiltrating to groundwater elevating the chloride contents. Hence
recharge is overestimated by this method for this particular catchment. Furthermore, part of the
chlorides released from effluents is flushed out by surface runoff.

Figure 46. Spatial distribution of groundwater recharge

Despite the fact that the method is simple and inexpensive, there are a number of uncertainties associated
with the method in estimating recharge. Given that input chloride concentrations can vary significantly
from site to site within a region of investigation, it is not surprising that CMB estimations are site specific
(Yongxin and Beekman, 2003). In most cases, the long-term average chloride in rainfall is not available.
Measured atmospheric input of chloride, often only short term records of chloride is assumed to be
representative for a long period. But an area of concern as rainfall and chloride deposition during the past
may be different from today. As discussed by Yongxin and Beekman (2003), other areas of concern include

54
the uncertainty in the measured chloride content of rainfall and rainfall amount. The largest uncertainty
associated with recharge estimation that utilises the chloride mass balance approach is the determination of
chloride concentration in the rainfall. Furthermore rainfall amount is generally difficult to measure, and is
highly variable. The absence of long-term rainfall quality data in the present study is one of the main
limiting factors affecting the accuracy of the method. Another uncertainty source for the chloride mass
balance approach is the sampling density and analysis accuracy of the chloride concentration of the
groundwater. As part of the fieldwork, samples from rainwater and groundwater were collected and
analysed for their chloride content that are utilised in the chloride mass balance method of recharge
estimation.

3.9.11.3. Baseflow separation: digital filter methods (recharge from baseflow)

Baseflow separation from hydrograph deals mainly with dividing the stream flow records into quick runoff
and baseflow components. Several base flow separation methods have been developed. One of the methods
is the recursive digital filter separation method. This method has been used in signal analysis and
processing to separate high frequency signal from low frequency signal (Lyne and Hollick, 1979). The
method involves in baseflow separation as high frequency waves which can be associated with direct
runoff and low frequency waves with baseflow (Eckhardt, 2005). Thus, the following digital filter
equations, obtained from Automated Baseflow Separation for Canadian Datasets (ABSCAN) software,
have been used in baseflow separation from total river flow.
Eckhardt (2005) presented general formulation of a two-parameter filter in which the baseflow index,
BFImax and the filter parameters Į need to be determined.

(1 − BFI max )aQb (i −1) + (1 − a ) BFI max Qi

Qb ( i ) = (11)
1 − aBFI max

Where Į refers to the groundwater recession constant and BFImax sets the maximum value of the baseflow
index BFI which is the long term ratio of baseflow to total streamflow. Qb(i) is the baseflow at time step i,
Qb(i-1) is the baseflow at the previous time step i-1, Qi is the stream flow at time step i, BFImax is the
maximum value of the baseflow index that can be measured. In fact it is a subjective parameter that needs
to be determined based on the type of catchment. Eckhardt (2005) suggested values for his parameter
BFImax based on the results obtained in his work in different catchments. He mentioned that BFImax
values about 0.80 for perennial streams with porous aquifers, 0.50 for ephemeral streams with porous
aquifers, and about 0.25 for perennial streams with hard rock aquifers, but he noted that this should be
further studied in order to determine it especially using tracer experiments.

In this study, a baseflow index value BFImax of 0.25 was used as a predicted value in accordance with the
Eckhardt’s assignment for perennial streams with hard rock aquifers since the Akaki catchment is
characterised mostly by fractured volcanic aquifers. Besides that recursive digital filter with filter
parameter of 0.995 were used to separate daily stream flow data into direct runoff and base flow. Based on
the Eckhardt filter method, the total annual river flow of 540 MCM (360 mm) and total annual baseflow of
95 MCM (63 mm) has been determined by current study. This means that according to this method, the
total annual recharge of the Akaki catchment is about 95 MCM (63 mm). (See Figure 47 below)

55
 18

16
Flow (MCM/day).

Total flow MCM/d

14 baseflow MCM/d

12

10

0
1
212
423
634
845
1056
1267
1478
1689
1900
2111
2322
2533
2744
2955
3166
3377
3588
3799
4010
4221
4432
4643
4854
5065
No. of days
Figure 47. Baseflow simulation from total flow in Akaki catchment

3.9.11.4. Recharge estimation using semi-distributed water balance model

With the aim of estimating groundwater recharge at monthly time step, a semi-distributed water balance
model was developed by AWSA (2000). The effect of urbanisation on the hydrological processes is taken
care of by considering a portion of the ground surface to be impervious. Rain that falls on impervious
surface is assumed to be completely converted into runoff after fulfilling the initial minor losses. Thus, the
impervious portion of the catchment area doesn’t contribute to both evapotranspiration and infiltration
processes. The structure of the model for pervious portion of the catchment is shown schematically in
figure 48. In the figure 48, BF stands for base flow, SRO for surface runoff, and RIF for interflow.
The model has three parameters:
SMAX = the maximum water holding capacity of the unsaturated soil
B = the portion of the free water in the soil zone that becomes recharge
CG = baseflow recession constant
The runoff coefficient is considered to be a function of the soil moisture state. Whatever portion of the
rainfall that infiltrates into the soil, it will first satisfy the soil moisture deficit. The gravity water in the soil
store, which is soil moisture in excess of the maximum amount of tension water, SMAX, will be
partitioned into groundwater recharge and interflow (or subsurface flow) based on the value of the
parameter B. That is,
IF(SM>SMAX) THEN
RECHARGE (RCH) = B * (SM-SMAX)
INTERFLOW (RIF) = SM-SMAX-RCH
ELSE
RCHARGE (RCH)=0
INTERFLOW(RIF)=0
ENDIF
It is assumed that the aquifer is always feeding the river in the upper part of the catchment in the form of
baseflow, which is estimated, considering the groundwater store to act as a single linear reservoir, as:

56
BF=BF0*CG + RCH*(1-CG)
Where, BF0= base flow in the previous month

Figure 48. Schematic diagram of the water balance model

The Akaki catchment area up to Aba Samuel hydropower plant is subdivided into nine sub-catchments in
order to account for spatial variation of input variables such as rainfall and catchment characteristics (or
model parameters). The potential evapotranspiration measured at Addis Ababa Observatory is assumed to
apply for the entire catchment, as there is no other Meteorological Station in the region where
evapotranspiration is measured. Furthermore, the parameter values are assumed not to vary from one sub-
catchment to another for this particular version of the model.
The model is then applied for each sub-catchment and the runoff generated over each sub-catchment is
accumulated to yield the simulated flow volume at Aba Samuel. The monthly flow volumes observed at
Akaki Bridge have been transposed to the catchment’s outlet at Aba Samuel by area ratio method, and
these derived values are taken as observed values for the purpose of model calibration and verification. The
model is calibrated against eleven years (1985 – 1995) monthly runoff data by minimizing the sum of the
square of differences between observed and simulated values, and then verified by using data from an
independent period of three years (1996 – 1998). Table 4 gives the optimized values of parameters and the
statistical measure of model efficiency in the calibration and verification periods.

Knowledge in groundwater recharge of the catchment is one of the critical parameters required in water
balance calculations, which in turn provide valuable information in long-term resource utilization.
Based on the observation of the large spatial and temporal variability in climatic condition as well as soil
and land use hetrogeniety, AAWSA (2000) developed a semi distributed monthly water balance model
rather than the common empirical methods in a recharge estimation. The recharge estimation has been one
of the key factors in assessing the potential of the catchment. Therefore, the assumptions made, the method
of calibration and verification, and the results obtained of the study are summarized below.

Assumptions made
• Recharge area: all areas except where low permeable lacustrine clay (Aba Samuel lake, the
wellfield, Dukem plain and bole areas) and thick clay soils (Sendafa and Alemgena areas) are
widely present. (See Appendix 12)

57
• Runoff: rain that falls on the impervious surface was assumed to be completely converted into
runoff after fulfilling initial minor losses (no infiltration process).
• It was assumed that the aquifer was always feeding the river in the upper part of catchment in the
form of base flow, which was estimated, considering the groundwater store to act as a single linear
reservoir.
• The potential evapotranspiration measured at Addis Ababa Observatory was assumed to apply for
the entire catchment, as there was no other meteorological station in the region where
evapotranspiration was measured.
• Further more, the parameter values were assumed not to vary from one sub catchment to other for
this particular case.
Modelling
• The Akaki sub-catchment area up to Aba Samuel hydropower plant (see Appendix 12 for map) was
subdivided in to nine sub-catchments in order to account for the spatial variation of input variables
such as rain and catchment characteristics (model parameters). The model also accounts for
transfer of water from one part of the catchment to another and interaction of different land uses
with the hydrological processes.
• The model was then applied for each sub-catchment and runoff generated over each sub-catchment
was accumulated to yield the simulated flow volume at Aba Samuel.
• The model took aerial average monthly rainfall and potential evapotranspiration for each
subcatchment and gives out monthly runoff at the outlet of the catchment and groundwater
recharge.

Calibration
• Automatic calibration of SMAX, CG and B is done by means of the simplex method (Press et al,
1986)
• The monthly flow volumes observed at Akaki bridge have been transposed to the catchment outlet
by area ratio method, and these derived values were taken as observed values for the purpose of
model calibration and verification.
• The model was calibrated against eleven years (1985 – 1995) of monthly runoff data by least
square calculations between observed and simulated values, and then verified by using data from
an independent period of three years (1996 – 1998)
The efficiency of the model in terms of simulating the observed runoff values was evaluated by the
Nash-Sutcliffe coefficient criteria. Efficiencies of 88 and 79 are obtained for the calibration and
verification periods respectively.
Results and accomplishments
• Once the monthly recharge series for each sub catchment was generated, the mean annual recharge
value for each sub-catchment was computed by summing the whole series and then averaging it
over the number of sub-catchments using weighted averages.
• The over all mean annual recharge (infiltration) rate in the Akaki river catchment has been
estimated to about 87 mm/year. This figure is higher than 51 mm/yr originally found by AAWSA
(2000). But this seemed to be caused by small errors in the FORTRAN code provided by AAWSA
(2000).

58
 Nash-Sutcliffe coefficient
Nash-Sutcliffe coefficient measures the efficiency of the model by relating the goodness-of-fit of the
model to the variance of the measured data, Nash-Sutcliffe efficiencies can range from -’ to An
efficiency of 1 corresponds to a perfect match of modelled discharge to the observed data. An efficiency
of 0 indicates that the model predictions are as accurate as the mean of the observed data, whereas an
efficiency less than zero (-’<E<0) occurs when the observed mean is a better predictor than the model.
Besides, due to frequent use of this Nash-Sutcliffe coefficient, it is known that when values between 0.6
and 0.8 are generated, the model performs reasonably. Values between 0.8 and 0.9 tells that the model
performs very good and values between 0.9 and 1 indicates that the model performs extremely well
(Deckers 2006).

¦ (Q − Q obs (i ) )
n
2
sim(i)
( 12)
R 2 = 1− i =1
2
n
§ ___
·
¦ ¨© Q
i =1
obs (i ) − Q obs (i ) ¸
¹
___
Q Q Q
Where: R : Nash-Sutcliffe coefficient, Sim (i ) : simulated flow, obs (i ) : observed flow and obs (i ) :
2

average of observed flow. The value of coefficients for both calibration and verification phases are shown
in the Table 10.

Table 10. Summary of results of the monthly water balance model

Optimized parameter values Model efficiency (R^2) %
SMAX (mm) B (-) CG (-) Calibration Verification
292.297 0.99 0.418 87.72 79.14

Figure 49. Compares graphically the observed and simulated monthly flow volumes at Aba Samuel for the
entire length of record. As can be seen from this figure the two series match reasonably well particularly in
the calibration period.
600

Calibration Verification
500
observed
simulated
Flow volumes (MCM)

400

300

200

100

0
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106 113 120 127 134 141 148 155 162
Month of record

Figure 49. Monthly Flow Volumes at Aba Samuel

Once the monthly recharge series is generated, the mean annual recharge value for each sub-catchment is
computed by summing the whole series and then averaging it over the number of years of record. It should

59
be noted that the mean monthly and annual recharge values reported in Table 11 were obtained after
accounting for the portion of areas of Black Cotton Soil in each subcatchment, which are considered to
have no contribution in recharging the aquifer. The figure 50 yields a mean annual recharge value of 87
mm/year for the entire catchment (See Appendix 14 for map).
Table 11. Average monthly and annual recharge (mm) for each subcatchment.
Subcatchment Jan Feb Mar. Apr. May June July Aug. Sep. Oct Nov. Dec Annual Area
sub1 0 0 0 0 0 0 37.5 61.4 0 0 0 0 98.9 167.3
sub2 0 0 0 0 0 0 46.4 60.9 0 0 0 0 107.3 223.5
sub3 0 0 0 0 0 0 16.2 52.7 1.1 0 0 0 70.1 115.2
sub4 0 0 0 0 0 2 31.2 85.0 4 0 0 0 122.3 242.1
sub5 0 0 0 0 0 1.7 19.2 71.8 2.5 0 0 0 95.2 252.7
sub6 0 0 0 0 0 0 7.6 52.6 0 0 0 0 60.2 250.6
sub7 0 0 0 0 0 0 18.0 55.9 4.2 0 0 0 78.1 36.0
sub8 0 0 0 0 0 0 8.7 56.9 4.9 0 0 0 70.4 63.2
sub9 0 0 0 0 0 0 3.7 33.2 0.5 0 0 0 37.4 146.0

350
Flows)MCM)

300

250

200

150

100

50

0
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Time (years)

Calculated flow Calculated baseflow

Figure 50. Calculated total flow and base flow at Aba Samuel
250.00
Flow (MCM).

200.00
baseflow
recharge

150.00

100.00

50.00

0.00
1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

Time (years)
Figure 51. Relation between base flow and recharge with time

60
 700

600
monthly rain (MCM/month)

500

400

300

200

100

0
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
time (yr)
Figure 52. Monthly rainfall (1985 – 1999)

The following are the results of the semi distributed water balance model of the Akaki catchment:
ƒ The average calculated flow at Aba Samuel is 492 MCM/yr (328.5 mm/yr)
ƒ Recharge or base flow in the area is calculated as 130 MCM/yr (87mm/yr)

Note: AWSA (2000) has done this model and found recharge of 51 mm/year. The approach taken by
(AAWSA, 2000) seem to be reasonable, however as discussed later in the following section the exclusion
of significant part of the catchment as totally impermeable proves to be unrealistic. Besides that errors in
the water balance are corrected. Hence, the mean recharge rate of the catchment is believed to be greater
than 51mm, in which case even the estimate of Tehal (1991), which is 63 mm may be a conservative one.

Urbanization results in land surface impermeabilization and reduces the direct infilteration of excess
rainfall, and hence accelerates direct runoff. This surface impermeabilization includes construction of roofs
and paved areas such as major highways, minor roads, parking lots, industrial areas, air port, etc. since this
fraction of impervious surface, due to urbanization, within the urban catchment reduces groundwater
replenishment, it is deducted from the total urban catchment contributing to recharge. For the city of Addis
ababa, in which informal settlements are progressively consolidated in to urban center, about 25 % of the
city is assumed to be impervious surface which does not directly contribute to the direct natural recharge.
Similarly, Lerner et al. (1990) suggests 75% of natural recharge for urban areas with rainfall greater than
1000 mm/yr which is the case to the study area.

Conclusions on the recharge of the catchment

Recharge to hydrogeologic units occur through natural processes and antropogenic activities, including:
• Direct recharge from precipitation through out the catchment, and surface and subsurface runoff
occurring primarily in the Furi, Wechecha and Intoto mountains and hills located along the western,
Northwestern, northern and northeastern boundaries of the Akaki catchment.
• Indirect recharge because of the leakage from sewers and water mains in the Addis Ababa city sector of
the catchment

61
 • Indirect recharge as the result of leakage from reservoirs and streams, traced especially in groundwater
within the southern part of the catchment.

Three surface water reservoirs (Dire, Gefersa, and Legedadi) give 80% of water supply to the city of addis
Ababa which in total of 175000 m3/day. Total well abstraction from the catchment including Akaki
wellfield is 23 MCM/yr (may be larger now). Hydrogeological investigation carried out in Akaki area by
AAWSA-THAL(1992) as cited in Tamiru (2001) showed that from the total water supplied to AA, about
70% returns as sewerage to the rivers and 60% of the returned flow has an outlet through big Akaki river
and the remaining 40% joins little Akaki river. In the study area there is a significant contribution from
sewerage that passes through drainage system into nearby streams. Ward et. al (1990) as cited in Tamiru
(2001) noticed that the discharge of effluents in to the stream channels and abstraction of water from the
stream channel may represent a very large percentage part of the natural flow and must therefore, be taken
in to account in the integrated water balance analysis. “Approximate indicative” conceptual chloride mass
balance in the catchment” is written as below)
• So chloride mass balance equation in the catchment will be modified in to the following equation

C p p = C ET + CuU + C gw A + CU DR + C R R + C gw B (13)

Where:
Cp = Chloride in precipitation (mg/l)
P = Precipitation (mm/yr)
CET = Chloride content on evapotranspired water (mg/l)
ET = Evapotranspiration
Cu = Chloride of Urban water supply (dams)
U = Urban water supply
Cgw = Harmonic mean of groundwater chloride (mg/l)
DR = Direcr runoff
B = Baseflow
R = Return flow (70 % of the urban water supply, U).

Figure 53. Groundwater surface interactions in the study area.

62
4. Calibration and execution of Akaki catchment
model
The calibration process involved adjusting of the inputs of the model such as parameters, boundary
conditions and sinks or stresses etc. in order to make a good match between the simulated and observed
state variables. The model to be a good representative of the real world, the difference between the
simulated and observed state variables should be as small as possible. This procedure requires the
calibration target which is referred to as goodness of fit criterion (Rientjes, 2007). The calibration target is
commonly defined as calibration value with its associated error (Rientjes, 2007). The error in the
calibration is determined by various aspects like accuracy of measurements, complexity of the system being
modelled and the applied model resolution. With these challenges, setting the calibration criterion is a
tedious task. Besides the real situation of the field, the calibration was done manually by trial and error
method until the minimum difference between the simulated and observed groundwater heads was
obtained. The changed parameters are hydraulic conductivity and recharge flux. The scatter plot of
measured heads against simulated heads was produced, and shows good results.

4.1. Calibration of the model

Calibration of the model requires that field conditions at a site to be properly characterized. Lack of proper
characterization may result in a model that is calibrated to a set of conditions, which are not representative
of actual field conditions (Anderson and Woessner, 1992).

4.2. Calibration target and uncertainty

Before starting calibration process in modelling, the selection of calibration target is required. Hydraulic
heads obtained from groundwater level measurement in the Akaki catchment were used as calibration
values. The calibration target was to match simulated hydraulic heads by the model with observed head
values. Hydraulic heads (DWL) for calibration purpose were obtained from groundwater level
measurement of year 2004. Besides that, static water level records, measured during the drilling time of
wells in the catchment as there were no data collected in water level after completion of drilling activities,
were used to check the accuracy of the model with out imposing well abstraction to it. See chapter 3.

It should be clear that most of the measured hydraulic heads data are associated with errors due to the
following reasons as is usually the case in Ethiopia.
ƒ The static water level measurements were taken just after well completion.
ƒ There are no monitoring wells which are well distributed in the catchment.

The monitoring wells are only found in the wellfield and the water level observations are carried out on
the pumping wells only in other part of the catchment. Because of this reason, the observed head values
may not represent the actual water levels.

63
 Table 12. Model Calibration Performance Measures (Aquaterra consultant, 2000)

Item Performance Measure Criterion

1 Water balance Less than 1% for each stress period and
Difference between total inflow and total cumulatively for the entire simulation.
outflow, including changes in storage,
divided by total inflow or outflow,
expressed as a percentage.
2 Iteration residual error Iteration convergence criterion should be one
The calculated error term is the maximum to two orders of magnitude smaller than the
change in heads (for any node) between level of accuracy desired in the model head
successive iterations of the model. results. Commonly set in the order of
millimetres or centimetres.
3 Qualitative measures Subjective assessment of the goodness of fit
Patterns of groundwater flow (based on between modelled and measured groundwater
modeled contour plans of aquifer heads). level contour plans and hydrographs of bore
Patterns of aquifer response to variations water levels and surface flows. Justification
in hydrological stresses (hydrographs). for adopted model aquifer properties in
Distributions of model aquifer properties relation to measured ranges of values and
adopted to achieve calibration. associated non-uniqueness issues.
4 Quantitative measures Criteria should be selected from the list of
Statistical measures of the differences residual head statistics. Consistency between
between modelled and measured head modelled head values (in contour plans and
data. Mathematical and graphical scatter plots) and spot measurements from
comparisons between measured and monitoring bores. Comparison of simulated
simulated aquifer heads, and system and measured components of the water
flow components. budget, notably surface water flows,
groundwater abstractions &
evapotranspiration estimates.

4.2.1. Steady State Flow Calibration

This process aims at checking the overall coherence of the selected assumptions of the modelling and at
identifying suitable hydraulic conductivity and recharge values. After numerous trials, satisfactory
simulation result of the flow, the piezometric levels and hydraulic conductivity has been obtained. The
calibration process was done using trial and error method by varying mainly aquifer hydraulic conductivity
values and comparing calculated heads to those measured in wells.

The most effective calibration technique for the adjustment of the hydraulic conductivity field in the model
was to initially delineate fewer conductivity zones and then gradually increase the number of zones based
on the geology and permeability of the hydrogeologic unit in the area. Hydraulic conductivity was
continually adjusted during calibration according to the geology and permeability in each hydraulic
conductivity zone.
64
 Figure 54. Trial and error calibration procedures (Adapted from Anderson and Woessner, 1992)

4.2.2. Calibration results

The best fit results (Fig 55 & 56) were achieved when the model domain was divided into regions with
different hydraulic conductivity zones. Subdivisions into zones were done based on geology and
hydrogeology of the model domain. As the result of this, subdivision in to zones should be regarded as
indicative only and then fine tune the hydraulic conductivity values until the model results match with field
situations.
(m)

(m)

Figure 55. Scatter diagram of calculated and observed heads in the study area

65
 Heads (m) Observed V, Simulated heads
observed
2700 simulated
2650
2600
2550
2500
2450
2400
2350
2300
2250
2200
2150
2100
2050
2000
T4
B5

B7

A1

A2

A3

A4

A5

C1
C2

C3
4

8
3b

M b
1b

M 2
03

04
B1

B1

A1

A1

A1

A1
A1

A2

A2
TW
5

0
BH

BH

W
M

BH Name
.
Figure 56. Comparison between actually measured and model computed heads

For calibration purpose, SWL of the boreholes are not used since the catchment’s model is done with
significant amount of groundwater withdrawal. Because of this reason, dynamic water levels of bore holes
listed in the table 13 below are used for model calibration. For calibration the DWL values of only 27
boreholes in the catchment were used. Then, normalized static water levels of 364 wells were used in the
model with out well abstractions (external stresses) to see the accuracy of the model. See figure 57.

Table 13. Steady state flow Simulated vs. observed hydraulic heads (m.a.s.l)
Observed Simulated ABS
X Y
heads (m) heads (m) (obs-simu) (obs-sim) (Obs-sim)^2
481200 980000 2212.5 2206.352 6.148 6.148 37.797904
473566 978610 2056.5 2061.756 -5.256 5.256 27.625536
480900 978800 2097.4 2079.431 17.969 17.969 322.884961
473108 979851 2086.5 2075.587 10.913 10.913 119.093569
473069 979881 2072.93 2076.004 -3.074 3.074 9.449476
478713 974977 2049 2035.296 13.704 13.704 187.799616
476574 975607 2049.8 2034.148 15.652 15.652 244.985104
476454 976951 2018.9 2035.307 -16.407 16.407 269.189649
476523 976374 2016.7 2032.8 -16.1 16.1 259.21
476972 976152 2035.3 2030.833 4.467 4.467 19.954089
477185 975729 2036.8 2032.269 4.531 4.531 20.529961
473576 972821 2032.7 2039.876 -7.176 7.176 51.494976
473600 1001013 2566 2578.391 -12.391 12.391 153.536881
473700 1001012 2568 2580.722 -12.722 12.722 161.849284
474300 1001005 2576 2592.161 -16.161 16.161 261.177921
466200 1001008 2587.9 2573.239 14.661 14.661 214.944921
467200 1001017 2563 2570.711 -7.711 7.711 59.459521
468900 1001007 2537 2540.775 -3.775 3.775 14.250625
474000 999400 2490 2489.564 0.436 0.436 0.190096
475000 1001300 2620 2623.887 -3.887 3.887 15.108769
470000 996400 2432 2429.542 2.458 2.458 6.041764
469150 988720 2275 2267.528 7.472 7.472 55.830784
485115.1 993909 2370 2374.065 -4.065 4.065 16.524225
480667.4 989984 2243.36 2237.027 6.333 6.333 40.106889
458690.7 997661 2530.61 2543.086 -12.476 12.476 155.650576

66
 (m)

(m)

Figure 57. Scatter diagram of calculated & observed heads (m), when 364 bore holes are used.

4.3. Evaluation of calibration

The results of the calibration should be evaluated both qualitatively and quantitatively (Anderson &
Woessner, 1992). The calibrated results were evaluated based on the calibration target and assessment of
the mass balance of the system (table 12). Flow direction was determined based on the simulated head
distribution and comparison is made with the flow direction determined in the conceptual model and by
previous studies conducted by Tamiru (2001). A scatter plot of measured against simulated heads is
another way of showing the calibrated fit (Fig.51, 52, 53). The scatter plots are visually examined to see
whether points in a plot deviated from the straight line. Furthermore the calibrated model outputs were
evaluated by applying the three common ways of error quantifying methods (Mean error, Mean absolute
error and Root Mean Squared error).

0.5
ª1
RMSE= ¦in=1(h −h )
2º (11)
«¬n c o i»
¼

The mean error is the mean differences between observed heads and calculated heads

1 n
ME = ¦ (hc − ho ) i
n i =1
(12)

The mean absolute error (MAE) is the mean of the absolute value of the differences in observed heads and
calculated heads

67
 1 n
MAE = ¦ | (hc − ho ) i |
n i =1
(13)

Where, n is the number of observations, hc the calculated head (L) and ho is the observed head (L)

The calibration of the model will be evaluated by using all these objective functions giving emphasis to the
RMSE since it is best evaluation method as compared to the other two methods.

Table 14. Errors of the calibrated model

Evaluation function Error (m)
ME MAE RMSE
Value -0.39918 5.37964 8.05441

4.4. Groundwater balance

One of the most basic ways to quantitatively evaluate the movement of groundwater through an aquifer
system is through the water budget. The fundamental equation for a water budget is that the sum of inputs
minus the sum of outputs equals the change in storage of the groundwater system:

ΔS = Qin − Qout + R − E g − Qabs (14)

Where ΔS = change in groundwater storage

For a groundwater system of the Akaki catchment, inputs to the aquifer include direct recharge from
precipitation, indirect recharge from surface water runoff, groundwater inflow from outside the system
boundary (only true for the wellfield), or recharge from anthropogenic sources (Lundmark et al., 2007).
Groundwater outputs of the area include discharge as springs, discharge to surface water bodies, loss to the
atmosphere by evapotranspiration (Eg), groundwater outflow to outside the system boundary, and pumping
for domestic, agricultural and industrial uses. It is summarized for steady state as:

R + Qin = Qout + E g + Qabs (15)

R = Groundwater recharge from rainfall

Qin = Lateral groundwater inflow (zero since no flow boundary from northern, western & eastern sides)
Qout = Groundwater outflow from outlet of the catchment, base flow and springs
Eg = Groundwater evapotranspiration (negligible) because of afro alpine climate.
Qabs = Flow from external groundwater sources and sinks (e.g. well abstractions)

Groundwater evapotranspiration reduces effective recharge or discards recharge completely causing a

lowering of the groundwater table in extreme cases (Lubczynski, 1997). Evapotranspiration comprises of
the surface evaporation, evapotranspiration from the unsaturated zone, and evapotranspiration from
saturated groundwater zone. It must however be emphasized that no method has been developed yet to
determine groundwater evapotranspiration correctly. Most deep rooting trees, capable of taking
groundwater from depth, have been cut over the last decades. Besides that there is no data regarding

68
groundwater evapotranspiration and difficult to quantify with limited time. Because of these reasons, it is
assumed as negligible. The water balance is established based on the modelling water budget tool in
Modflow.

Table 15. Water balance in Steady state flow obtained through Akaki catchment model calibration
2
area= 1500 km
3 3 3
m /day m /day m /day MCM/yr mm
Flow term IN OUT IN-OUT IN-OUT IN-OUT
Constant head 16791 104099 -87308 -32 -21
Wells 0 62763 -62763 -23 -15
Drains 0 11747 -11747 -4 -3
Recharge 432672 0 432672 158 105
ET 0 0 0 0 0
River leakage 2992 252912 -249920 -91 -61
Head dependent boundaries 0 20933 -20933 -8 -5

SUM 452455 452454 1 0 0

From this water balance, the total average base flow obtained from the model is 92 MCM/yr
or 62 mm/yr. From water balance model described in chapter 3, it is found base flow and recharge value of
130 MCM/yr or 87 mm/yr. using digital filter method, the base flow and recharge values are roughly 95
MCM/yr (63mm/yr) indicating good agreement between the models.
The evaluation of the calibrated model result shows that:
ƒ Most of the simulated heads were within the pre-established calibration target.
ƒ Water balance discrepancy was zero.
ƒ The overall results of the groundwater model are comparable with the measured well data and in
agreement with conceptual model.
The measure of errors evaluated by ME, MAE and RMSE are in the acceptable range according to the pre-
determined error criteria.
Though the overall result of the model was comparable with the measured well data, few observations
which are not uniformly distributed over the model domain are utilised in the calibration process. Ideally
calibration values should be measured at a large number of points uniformly distributed over the model
domains. Thus it is not possible to conclude that the calibration is accurate by only quantifying the errors
using ME, MAE and RMSE with out considering the distribution of the residuals.

4.5. Groundwater outputs (in water balance)

The aquifers have three major outputs:

4.5.1. Springs
• Springs in Aba Samuel Gorge, which is due to sideways intersection of the piezometric surface with
the gorge. About 10 springs are reported in the gorge. The yield of the seven springs is estimated by
(AAWSA, 2000) and a collective discharge between about 30 l/s (2592 m3/day) and 60 l/s (5184
m3/day) considered to be reasonable.
• Fanta Spring, which is being used for water supply of Akaki town. The spring, which is capped, for
the water supply has a recorded yield (from Akaki Water Supply Water Meter) of 13 l/s (1123.2
m3/day) to 22 l/s (1901 m3/day) with average of about 18 l/s (1555.2 m3/day). However, its yield is
over 22 l/s (1901 m3/day) including the non-capped parts (see water balance of Akaki wellfield model).

69
• Other springs, there are a number of other small springs over the study area whose discharges are not
measured frequently. For these springs a total of 30 l/s (2592 m3/day) is assumed excluding the hot
springs of central Addis Ababa.

4.5.2. Base flow to the rivers

Essentially part of the flow of big Akaki River comes from groundwater discharge. In dry season this
contribution gives the major part of the observed rate. According to 2008 model, the base flow to the rivers
from groundwater is 252912 m3/day, which is 92 MCM/yr.

4.5.3. Abstraction of the wells

As described above the total groundwater abstraction with boreholes is estimated to be about 136986
m3/day) which is approximately 23 MCM/yr (15mm/yr). For a volcanic aquifer such production rates are
high. Production over 20 % of the estimated recharge in a hard rock aquifer may not be advisable for a
sustainable use, because the estimated recharge might not be available as it can get access to be conveyed
to deeper zones or other areas following fracture.

4.6. Model validation

In model validation, the normal procedure is to define a set of measurements or observations of system
variables, where part is used for model calibration and the remaining part is used for model validation.
Model validation was not possible here because the data set was too short.

4.7. Uncertainty of the model calibration and model limitations

Uncertainty in hydrologic modelling may be due to model conceptualization, input parameters and/or
inherited in natural processes. Simply, model uncertainty arises from incomplete understanding of the
system being modeled or inability to accurately reproduce hydrological processes with mathematical and
statistical techniques (Das & Lewis, 2007). The study area is complex in terms of geological setting and
hydrometeorological processes which give rise to significant heterogeneities and anisotropy.

The built model of the present research is associated with a number of uncertainties. First of all the
hydrogeological heterogeneity caused difficulties in the conceptual simplification of the field condition.
Despite the complex and heterogeneous nature of the aquifer system, assumptions and simplifications were
made during the conceptualization of the system. Definitely uncertainties will be introduced as a result.
Various forms of heterogeneity in the porous media properties can be very different from the fluid flow
behaviour in the individual zones (Das & Lewis, 2007). There are generally few locations where
observations are available, and the geological structure of the aquifer is only partially known. The main
constraints in the process were data gaps and the poor quality of the available data. Important data such as
screen length and aquifer thickness are not well documented.

Another area of uncertainty is resulting from defining the boundary conditions of the model domain. The
boundary conditions were defined based on the surface physical features such as geology, rivers, dams and
surface water divides. Finally, the area is conceptualized as a single layer which can probably simplify the
system much.

70
5. Akaki wellfield flow modelling

5.1. Introduction
The Akaki wellfield is situated in the lower part of the Akaki River catchment within the drainage basin of
Dongora and Keta which join to form Sekelo, which in turn flows into the big Akaki river. The drainage in
the area in general flows southwesterly to Aba Samuael Lake. Most of the streams in the area are
intermittent except the Akaki River. Dendritic drainage patterns are apparent in the area. Although the
lithology is mainly volcanic in origin, with different successive lava flows, the scoraceous lavas have a
very high storage and transmissivity. For this reason, a wellfield was developed here with a large number
of wells, clustered closely together. (See figures 4 and 5)

5.2. Hydrogeology of the wellfield

5.2.1. Geology

Lithologic units in the wellfield

Volcanic rocks dominate the study area with subordinate alluvial sediments. The volcanic rocks are the
lower basaltic flows and younger basaltic scoria and lava (Aynalam, 1999). The lower basalt flows
constitute the oldest outcropping rock unit and the alluvial sediments along the Akaki River and Sekelo
stream form the youngest unit. These lower basalt flows are exposed in the western part of the study area.
The logs show intercalations of massive basalt, scoraceous basalt and pyroclastic rocks such as scoria and
tuff (Aynelam, 1999). The eastern part of the study area exposes younger basaltic rocks dominated by
scoria cones and associated flows. The scoria cones are aligned along northeast-southwest direction,
parallel to the trend of rift faults. These rocks are in places covered by recent alluvial sediments. The
lithology at the wellfield is extremely variable. Mixture of alluvial and lacustrine materials such as sand,
clay, gravel, volcanic ash and tuffs are variably found at certain depths. In general, the thickest scoria
deposits are located in the EP wells series of the Akaki town water supply situated at the north-eastern part
of the wel field (See Fig. 58).

Structures and weak zones

As stated previously, Addis Ababa is situated at the western margin of the Main Ethiopian Rift. The rocks
are subjected to rift tectonics, as manifested in a number of fault systems having a general trend of the rift
system (NE-SW). There are also some faults and lineaments oriented E-W, N-S, and NW-SE. Some of the
basaltic lava and cinder cones concentrated along southeast to northeast of the wellfield likely erupted
following the NE-SW trending fault systems. A relatively dense fault network and lineaments is observed
in the wellfield (Fig. 58 and 59). Field investigations by AAWSA et al.,(2000) cited in Gizaw (2002)
suggested that micro-structures such as fissures, fractures, conduits and joints are abundant at the outcrops
on the mountain side and deep cut river sides in the vicinity of Addis Ababa which play a key role in
facilitating groundwater recharge and contaminant transport. Parts of the Akaki river bed are following the
main fault line. This also has its own implication in contaminant percolation at the river bed and migration
down the aquifer.

71
 N
AI
M
DO
L
DE
O
IM
AK
AK

Figure 58. Geological map of the study area (modified after AG consult, 2004).

Figure 59. Dense fracture traces in the wellfield area (modified after Tamiru et al., 2005).

72
Subsurface geology from geological and geophysical logs
Lithological logs were used to classify aquifer media, type of vadose zone and depth of soil profiles. An
attempt has been made to evaluate the geological log, and resistivity log of the boreholes in the area
(Appendix 1 and 2). Since resistivity logging is only possible below the static water level, the evaluation of
the upper parts of the borehole depends solely on the geologic log. The geologic logs of the study area
indicate that the major formation of the aquifer is basaltic in composition, while the water quality analysis
revealed that the water is generally fresh. Therefore, the shape of the resistivity log curves depends mainly
on the degree of fracturing and presence or absence of water. Correlation was found to be difficult due to
the lenticular nature of the units, rapid lateral changes within units, and variable dips, due to different
centers of volcanic activity depositing materials in different places in various periods. Since rocks of
various ages are distributed in the study area aquifer characterization becomes a difficult and complex task.

5.2.2. Regional and local groundwater flow directions

The elevations of water level in boreholes are used to determine the general direction of groundwater flow
in the study area. In general, the groundwater movement is sub-parallel to the surface water flow direction
and more or less controlled by the topography of the area. The piezometric surface constructed from
groundwater point inventory made during previous studies showed that the general groundwater flow
direction in Addis Ababa is from north to south in the upper & central part and towards south & south-east
in the lower parts of the catchment. AAWSA (1998) as cited in AAWSA (2000) assumed local
groundwater flow direction in the wellfield is from NE towards the SW where natural springs exist in the
Aba Samuel Gorge.
Analysis of water table data has shown that on relatively regional scale the groundwater flows from the
water divide to the discharge area on the river valleys. Convex contour lines at the northeastern and
northwestern corners of the area (Fig. 69) indicate regions of groundwater recharge, while concave contour
lines at centre and along Akaki River are associated with groundwater discharge areas. The regional
groundwater flow direction is therefore, from the north to the wellfield and Akaki River. However, because
of limitations of the water level data, particularly in the southern part of the study area, it is not possible to
depict the local groundwater flow directions in this area accurately. The flow lines, sketched perpendicular
to the contour lines, show the direction of groundwater flow (See Figure 69) Shallow local patterns of
groundwater flow near surface water are emphasized in this study, as shallow aquifers are more susceptible
to contamination from anthropogenic sources.

5.2.3. Groundwater recharge and discharge conditions

In developing a conceptual model of a flow system, it is important to consider the topographic setting.
Topographically higher areas are typically zones of intake or recharge, while topographically lower areas
are areas of discharge. In most areas, the volcanic aquifers show locally confined, mostly semi-confined to
unconfined nature. The recharge to the groundwater which takes place within the Akaki catchment to the
north of Akaki Bridge is considered contributing to the base flow. Water enters a confined aquifer in an
area where the confining beds rise and the aquifer is exposed to surface (recharge areas), and also enter by
leakage through fractures and pores of the confining beds.
In the study area, water enters into aquifers from natural recharge areas such as at a number of scoria cones
where bed rocks are exposed and where clay and black cotton soil coverage is thin. Eventually there is
water entering to the study area from the upper boundary as influx to it. Groundwater also flows towards

73
discharge areas, where it flows out as springs, seepage zones or it may be tapped by a number of wells in
the wellfield or drainage systems. The young patches of basaltic scoria such as Indode, Mt. Bilbilo,
Mt.Guji, Gerado, Gara Bushu, Dengora Chefe, EHA Quarry and Tulu Dimtu are the main local recharge
areas.

5.2.4. Water level

Depth to water level in each well was obtained by subtracting the water level elevation from the ground
surface elevation. Depending on the hydrogeological setting, the depth to water does not necessarily
coincide with the Static Water Level (SWL). Therefore, to evaluate such conditions the depth to water is
determined by the type of aquifer (confined or unconfined) and information extracted from well data.
Where confined aquifers were identified (Kality area), the corresponding SWL of that well was excluded
from mapping depth to water. The groundwater level around Fanta and Kality are shallow in the range of 1
m to 34 m below ground surface. Comparisons of ground elevations and static water elevation at respective
wells of the area are shown in Fig. 60. The potentiometric surface indicates that the groundwater is in
connection with the surface water of Big & Little Akaki Rivers north of Akaki Bridge.
2250.00

2200.00

2150.00
Elevation (masl)

2100.00

2050.00

2000.00

1950.00

1900.00

1850.00
Ethiopian Iron&Steel
Indo-Europian-3

Adwa Elour Mlill

Aka.Keb.06 Kilento
Kality Food BH-1

Kality Airforce-1

Ethio-Metal Meal-1
AAWSA Kality Well
Kality Soap&

Akaki Metal

Akaki Metal
NET-SA Plc, Kality
Kality MA

Akaki Water EP-4

Akaki Water EP-7

Water III BH25-2

Akaki Beverly
Water III MW02
Water III BH01

Water III BH3b

Water III BH06

Water III BH09

Water III BH12

Water III BH16

Water III BH19

Water III BH22

Water III Tw-T5

Sidamo Awash
IAEA (P3)
TW EP-1

AAWSA F3

Locality

Ground Elevation (masl) (Z) Elevation of SWL (masl)

Figure 60. Comparison of ground surface elevation and normalized SWL in the wellfield.

Figure 61. Conceptual model of Akaki wellfield.

74
5.3. Groundwater model of Akaki wellfield
A conceptual model is a simplified representation of the essential features of the physical hydrogeological
system, and its hydrological behaviour, to an adequate degree of detail.

5.3.1. Spatial discretization of model grid

In MODFLOW, an aquifer system is replaced by a discretized domain consisting of an array of nodes at
which hydraulic heads are calculated and associated in finite difference blocks (cells). The model area
encompasses the limits of the local flow system of the Sekelo and Keta streams and extended up to Mt.
Bilbilo and Guji to the south. The model spans an area of 197.3 km². The model grid consists of 44
columns and 74 rows. It consists of 3256 cells with a regular grid spacing of 400 m x 200 m. The
geographic boundaries of the model grid were determined by using a DTM. A finite-difference grid
superimposed over a 197.3 Km2 area was designed and constructed based on the simplification of a
conceptual model representing the physical properties of the groundwater system.

5.3.2. Input parameters of the wellfield model domain

The spatial input parameters are initial hydraulic head, horizontal hydraulic conductivity, recharge, top of
aquifer, bottom of aquifer, and effective porosity. Therefore, values were assigned to each active grid cell,
based on its location within the study area. Top of aquifer elevation was assigned based on the static water
level of the area and bottom elevation of the aquifer was obtained by subtracting one hundred meters from
the top of the aquifer

Figure 62. Block diagram shows conceptual model of Akaki wellfield

75
 Figure 63. Input parameters of wellfield model

5.3.3. Model boundary conditions

Constant head is fixed to Abba Samuel Lake since its level does not much fluctuate with time. The base of
the volcanic aquifer is the basement volcanic rock where groundwater flow is negligible. Since no flow is
assumed to the bottom of the volcanic aquifer, it is taken as a specified-flux (no-flow) boundary. No flow
boundary has been assumed in eastern boundaries of the study area (Fig 65). Head dependent boundaries
are assigned to the northern, North western and southern part to simulate incoming and outgoing fluxes in
the model domain. By default and convention the area outside the model domain is set to inactive.

1.2
1 Aba Samuel Lake level
lake level (m)

0.8
0.6
0.4
0.2
0
1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

years
Figure 64. Aba Samuel lake level (MOWR, 20007)

76
5.3.4. Surface-groundwater interaction
Surface-groundwater interaction can form a critical component of the water budget, as well as an essential
feature of the conceptual model, and often forms the most complex and uncertain parts of a model. From a
groundwater perspective, it is commonly assumed that direct simulation of the unsaturated flow system is
not critical, and that leakage from surface water to groundwater occurs instantaneously. To simulate
groundwater–surface water interaction in the model domain river package, type of head dependent
boundary, is used. It is used to assess the gains to the river (base flow) and the loss of water from the river
towards aquifer (river leakage). In addition, constant head boundary at Aba Samuel lake is used to simulate
the groundwater and surface water interaction process.

Figure 65. Boundary conditions of the wellfield model

The unregulated well abstractions are some of the external stresses facing the Akaki wellfield. The total
well abstractions imposed to the model domain are long term average abstractions until to date. At the
wellfield, it is taken as long term average abstraction but at the northern, north western and north eastern
sides of the model domain, the abstraction records of 2004 are imposed. In this way, the total abstraction
given to model is 58604 m3/day.

77
 Table 16. Well abstractions applied to the model.
Layer Row Col Stress rate (m3/day) Well no.
1 7 16 -88 1
1 8 16 -88 2
1 9 18 -220 3
1 9 34 -150 4
1 11 30 -150 5
1 13 22 -137 6
1 13 28 -150 7
1 14 34 -547 8
1 15 13 -94 9
1 16 13 -94 10
1 17 19 -123 11
1 17 22 -125 12
1 17 30 -1901 13
1 18 22 -125 14
1 21 27 -2400 15
1 25 24 -137 16
1 27 25 -300 17
1 31 27 -136 18
1 33 28 -1340 19
1 33 30 -2500 20
1 33 31 -500 21
1 35 28 -5216 22
1 35 29 -7808 23
1 36 29 -5874 24
1 37 27 -6912 25
1 38 27 -4384 26
1 38 29 -1116 27
1 39 26 -6496 28
1 39 29 -3616 29
1 40 28 -1472 30
1 41 25 -860 31
1 41 26 -328 32
1 41 27 -2280 33
1 53 11 -549 34
1 55 16 -390 35

5.4. Model calibration

General
A graphical comparison between actually measured and model computed heads with and without pumping
scenario is shown in Figs 66, 67 and 68. After numerous trials, satisfactory simulation result of the flow,
the piezometric levels and hydraulic conductivity has been obtained

78
 2140
Heads (m).

2120
observed head
2100 simulated head

2080

2060

2040

2020

2000
B5
B6
B7
B9

B13
B14

Ep-1
Ep-3
B15

BH03b
BH04
BH05b
BH06
BH07
BH08
BH12
BH14
BH17
BH20
BH23
BH24
BH25-2
BH26
Ep-4
Ep-5
Ep-6
Ep-8
T1
T5

T2

T4

Borehole name

Figure 66. Wellfield steady state model calibration using static water levels with no abstractions

Total of 31. boreholes are used for model calibration and they are shown in the scatter diagram below.

Table 17. Errors of the calibrated model (unit: meter)

Evaluation functions ME MAE RMS
Values -0.154 0.713 0.768

Figure 67. Calculated and. observed heads (m) of the wellfield without well abstraction

79
 Table 18. Boreholes used for model calibration under non-pumping scenario.
BH_name observed head(masl) simulated head (masl) 0bs-simu ABS(0bs-simu) (0bs-simu)^2
B5 2075.8380 2074.9620 0.8760 0.8760 0.7674
B6 2076.7700 2077.3340 -0.5640 0.5640 0.3181
B7 2041.3250 2042.2090 -0.8840 0.8840 0.7815
B9 2124.0000 2123.4450 0.5550 0.5550 0.3080
T1 2075.8000 2074.9620 0.8380 0.8380 0.7022
T5 2124.0000 2123.4450 0.5550 0.5550 0.3080
B13 2082.7000 2083.2800 -0.5800 0.5800 0.3364
B14 2043.4000 2043.3790 0.0210 0.0210 0.0004
T2 2083.1630 2083.2800 -0.1170 0.1170 0.0137
Ep-1 2082.6000 2082.8390 -0.2390 0.2390 0.0571
Ep-3 2085.0000 2085.7720 -0.7720 0.7720 0.5960
B15 2050.0000 2049.5070 0.4930 0.4930 0.2430
T4 2050.2700 2049.3760 0.8940 0.8940 0.7992
BH03b 2029.4890 2030.4460 -0.9570 0.9570 0.9158
BH04 2030.2000 2030.4030 -0.2030 0.2030 0.0412
BH05b 2029.4000 2030.1910 -0.7910 0.7910 0.6257
BH06 2035.0000 2034.4960 0.5040 0.5040 0.2540
BH07 2031.4000 2032.3280 -0.9280 0.9280 0.8612
BH08 2029.8000 2030.6350 -0.8350 0.8350 0.6972
BH12 2032.2000 2031.4060 0.7940 0.7940 0.6304
BH14 2031.4500 2030.5480 0.9020 0.9020 0.8136
BH17 2029.8000 2030.5410 -0.7410 0.7410 0.5491
BH20 2030.5000 2031.2490 -0.7490 0.7490 0.5610
BH23 2031.9000 2032.8290 -0.9290 0.9290 0.8630
BH24 2029.7000 2030.5030 -0.8030 0.8030 0.6448
BH25-2 2029.7000 2030.3620 -0.6620 0.6620 0.4382
BH26 2029.4000 2030.3000 -0.9000 0.9000 0.8100
Ep-4 2038.0000 2037.1700 0.8300 0.8300 0.6889
Ep-5 2044.2000 2045.2540 -1.0540 1.0540 1.1109
Ep-6 2036.3500 2037.0840 -0.7340 0.7340 0.5388
Ep-8 2040.4260 2039.0140 1.4120 1.4120 1.9937

The acceptability of a calibration can be assessed by judging whether each of the performance measures
listed in Table 19 conforms to specified criteria. The RMSE of 1.1 meter is still much below the target but
figure 68 shows that under the circumstances the fit is quite good.

80
 2150.0

2125.0
heads (m)

2100.0

Simulated heads(masl)
2075.0 Observed heads(masl)

2050.0

2025.0

2000.0

1975.0

1950.0

BH03b

BH04

BH05b

BH06

BH07

BH08

BH14

BH17

BH20

BH23

BH24

BH25-2

BH26
B5

B6

B7

B9

B13

B14

Ep-1

Ep-3

B15

Ep-4

Ep-5

Ep-6

Ep-8
T1

T5

T2

T4

BH name

Figure 68. Observed & simulated heads of Akaki wellfield, calibrating with pumping senario.

Dynamic Normalized hydraulic head (DWL) valued collected during 2004 are used for Akaki wellfield
calibration with external stresses, well abstractions.

Table 19. Evaluation of calibration with pumping scenario

Evaluation of calibration
Eror measured (m) ME MAE RMSE

value 0.189 0.84 1.1

Table 20. Observed and simulated heads of Akaki wellfield with pumping scenario
BH Name x y Simulated heads(masl) Observed heads(masl) obs-sim (m)
B5 481200 980000 2075.82 2076.20 0.38
B6 470800 982900 2080.22 2079.30 -0.92
B7 473566 978610 2043.29 2043.33 0.04
B9 481600 982900 2125.35 2124.30 -1.05
T1 481200 980000 2077.22 2075.80 -1.42
T5 481600 982900 2125.35 2123.30 -2.05
B13 478400 981400 2081.01 2082.00 0.99
B14 480900 978800 2044.28 2044.40 0.12
T2 479400 981400 2081.01 2081.20 0.19
Ep-1 479340 981400 2080.50 2081.60 1.10
Ep-3 479740 981400 2082.17 2083.00 0.83
B15 473069 979881 2052.16 2051.80 -0.36
T4 473108 979851 2052.06 2052.30 0.24
BH03b 478713 974977 2002.74 2005.41 2.67
BH04 477992 975552 1999.47 2000.00 0.53
BH05b 476574 975607 2011.95 2011.40 -0.55
BH06 479696 976936 2008.20 2008.77 0.58
BH07 479405 976735 2000.93 2000.90 -0.03
BH08 479061 976370 1992.37 1995.12 2.75
BH14 478580 976051 1992.02 1993.00 0.98
BH17 478199 976361 1986.87 1986.90 0.03
BH20 477945 976985 1997.29 1997.29 0.00
BH23 477477 977216 2009.26 2009.05 -0.20
BH24 477330 976793 2004.27 2005.47 1.20
BH25-2 477162 976038 2004.83 2004.78 -0.05
BH26 477181 975680 2005.85 2007.12 1.28
Ep-4 479942 977322 2016.48 2017.60 1.12
Ep-5 478450 979950 2039.61 2040.00 0.39
Ep-6 479526 977468 2007.58 2005.75 -1.83
Ep-8 478998 977937 2020.76 2019.43 -1.33

81
Figure 69. Flow nets in the wellfield after calibrating the model with dynamic water level of 2004

Figure 70. Hydraulic conductivity values (m/day) used for wellfield calibration

82
5.5. Water budget of wellfield
For a groundwater system of Akaki wellfield, inputs include direct recharge from precipitation, indirect
recharge from surface water (Akaki river, Aba Samuel lake) (Qriv), groundwater inflow from upper part of
the system boundary (Qin). Groundwater outputs may include discharge as springs, discharge to surface
water bodies as baseflow to rivers, loss to the atmosphere by groundwater evapotranspiration (ET),
groundwater outflow to outside the system boundary (Qout), and well Abstruction for domestic, agricultural
and industrial uses. It is steady state flow which can be summarized as:

R + Qin +Qriv = Qout + Eg + Qbaseflow+Qwell + Q spring (18)

R = Groundwater recharge from rainfall

Qin = Lateral groundwater inflow from northern, western sides of the wellfield domain.
Qout = Groundwater outflow from out let of the wellfield, simulated by general head boundary
Eg = Groundwater evapotranspiration is negligible because of afro alpine climate.
Qwell = Ground water abstraction through wells
Qspring = Groundwater out flow to the surface via spring, which is simulated by drain package
Qbaseflow = Base flow to the Akaki river which is simulated by river package.

Table 21. Groundwater balance with out well abstraction in the wellfield
2
area= 197km
3 3 3
m /day m /day m /day MCM/yr mm/yr
Flow term IN OUT IN-OUT IN-OUT IN-OUT
Constant head 15208 15730 -521 -0.2 -1
Wells 0 0 0 0.0 0
Drains 0 2249 -2249 -0.8 -4
Recharge 42660 0 42660 15.6 79
ET 0 0 0 0.0 0
River leakage 1996 44753 -42756 -15.6 -79
Head dependent boundaries 21951 19084 2867 1.0 5

SUM 81816 81816 0 0.0 0.0

Table 22. Wellfield water balance with well abstraction, calibration by DWL(2004).
2
area= 197km
3 3 3
m /day m /day m /day MCM/yr mm/yr
Flow term IN OUT IN-OUT IN-OUT IN-OUT
Constant head 15691 15425 266 0.1 0
Wells 0 36170 -36170 -13.2 -67
Drains 0 1066 -1066 -0.4 -2
Recharge 42660 0 42660 15.6 79
ET 0 0 0 0.0 0
River leakage 9742 26496 -16754 -6.1 -31
Head dependent boundaries 24396 13265 11132 4.1 21

SUM 92489 92489 0 0.0 0.1

83
5.6. Sensitivity analysis
The overall performance of a groundwater model may be better analyzed through a sensitivity analysis of
its aquifer parameters. The sensitivity analysis allows the groundwater investigator to better understand the
system’s response to changing parameters. Sensitivity analyses were used to refine initial estimates of input
parameters during model calibration, and to determine which input parameters had the largest effect on
simulated head values after model calibration. If the model is sensitive to an input parameter, additional
data on that variable can help improve calibration. Increments and decrements of 25 percent were applied
to horizontal hydraulic conductivity, and rainfall recharge. The resulting hydraulic heads were then
compared with the observed hydraulic heads and mean average error; absolute average error and root mean
squared error were calculated for each parameter. Then the calculated average errors in the hydraulic heads
were plotted against the multiplying factors as shown in figures 71 and 72. It was found that slight changes
in either the aquifer hydraulic conductivity or slight changes in recharge rate affect dramatically the
distribution of hydraulic head throughout the area. The sensitivity plots show that the recharge generates
non-linear sensitive response while sensitivity towards hydraulic conductivity generates linear response.
The model is equally sensitive to both increase and decrease of hydraulic conductivity on the other hand;
the calibrated model is more sensitive to recharge fluxes increment than to recharge rate reduction.
40
Error of hydraulic head (m)..

RMSE MAE ME
30

20

10

0
-0.5 -0.25 0 0.25 0.5 0.75

-10

-20
Hydraulic conductivity change from calibrated value
Figure 71. Sensitivity plot of the calibrated model with respect to hydraulic conductivity
25
ME MAE RMSE
20
Error of hydraulic head (m)..

15

10

0
-0.5 -0.25 0 0.25 0.5 0.75
-5
Recharge change factor from calibrated value

Figure 72. Sensitivity plot of the calibrated model with respect to recharge

84
6. An advective solute transport modelling of
Akaki wellfield, PMPATH

6.1. General
PMPATH is an advective transport model running independently from PMWIN Pro (Chiang and
Kinzelbach, 1998). PMPATH retrieves the groundwater models and simulation result from PMWIN Pro
and MODFLOW. A semi-analytical particle-tracking scheme is used to calculate the groundwater paths
and travel times. Moreover, PMPATH provides various on-screen graphical options including head
contours, drawdown contours and velocity vectors for any selected model layer and time step. Both
forward and backward particle tracking are allowed for steady state and transient flow simulations. In this
respect, PMPATH can be used to simulate advective transport in groundwater to delineate contaminant
capture zones, and wellhead protection areas. PMPATH creates several output files including hydraulic
heads distribution, velocity field, and travel times of particles.

6.2. Hypothesis and assumption of the solute transport (PMPATH)

Water carrying contaminants (Fig. 73) may enter into the underlying porous aquifer from Tilu dimtu
quarries, highly polluted Akaki river (Gizaw, 2002) that drain the urban centre, and most streams of Sekelo
sub-basin into which most factories and industries directly release their effluent. Effluents are released both
under natural conditions where groundwater and surface water interact (loosing streams) in recharge areas
and exit at discharge areas (wet lands, water supply wells, lake and ponds), and under artificial conditions
where flow paths fall within the capture zone of wells under maximum pumping conditions.

Figure 73. Conceptualization of the process by which solutes transport by moving groundwater

It is assumed by PMPATH that fluid properties are homogeneous and that concentration changes do not
significantly affect the fluid density or viscosity and hence the fluid velocity. As the result of this, it is
assumed that pollutant moves at the same velocity as water. Given the initial position of a particle
anywhere in a cell, the coordinates of any other point along its path line within the cell, and the time of
travel between them, can be computed directly. For steady-state systems, the exit point for a particle
entering a cell at any arbitrary location can be computed in a single step. By following the particle as it
moves from cell to cell, it can be used to trace the path of a particle through any multidimensional flow
field generated from a block-centered finite-difference flow model.

85
6.3. Factors governing contaminant transport
Water pollutants tend to be removed or reduced in concentration with time and distance traveled (Tamiru et
al., 2005). Understanding the factors that govern contaminant transport particularly in this area is,
therefore, crucial. The rate of pollution attenuation depends on the geology, local hydrogeological
situations, geochemical processes and the type of pollutants. Moreover, mechanisms of pollution
attenuation include filtration, sorption, chemical processes, microbiological decomposition and dilution.
These factors affect contaminant migration in one or another, and are explained below.

6.3.1. Implication of geology for transport

The geology of the area including rate and extent of physical and chemical weathering of rocks, and the
density and orientations of the structures affect the rate of infiltration of polluted water. The moving water
leaches not only the free cations removed from the mineral structures but also transport species that enter
water as a consequence of chemical weathering. Thus, the weathered rocks of the area, associated
structures and their orientation have a facilitating effect in contaminant migration.

6.3.2. Hydrogeological suitability for transport

The groundwater circulation and the dispersion of pollutants depend on the hydrogeological characteristics
of the material such as porosity, permeability, and hydraulic conductivity. To identify the pathway and
final destination of pollutants, it is necessary to describe the earth materials with a particular reference to
their infiltration capacity (Tamiru, et al., 2005). Groundwater and contaminant movement as well as
accumulation in unconsolidated pyroclastic aquifers are determined by fragment size, sorting and the
degree of cementation of particles.

The infiltration capacity of water in the black cotton soils of the area is high at the beginning of the rainy
season through the cracks formed in the previous dry season, and reduces when the amount of precipitation
increases. As a consequence, contaminants that enter cracks of the black cotton soil during the dry seasons
will later move down with infiltrating water during the rainy season. On the other hand, when clay is not a
dominant constituent of the soil, there is relatively a constant infiltration of water in the rainy season
through the highly porous and permeable rocks of the area (Tamiru, et al., 2005).

Traditional water resources management treats surface and groundwater as separate entities. However, it is
apparent that the movement of water between surface and groundwater provides a major pathway for
chemical transfer between terrestrial and aquatic systems. Nearly all surface water features (streams, lakes,
reservoirs, wetlands, and estuaries) interact with groundwater in various ways. In many situations, surface
water bodies gain water and solutes from groundwater systems while in others surface water is a source of
groundwater recharge and causes change in groundwater quality (Fig 74). Pollution of surface water can
cause degradation of groundwater quality and conversely pollution of groundwater can degrade surface
water. Thus, effective water management requires a clear understanding of the linkage between surface and
groundwater at any point and time in a given hydrogeologic setting.

86
Figure 74. Rivers and groundwater connection; a gaining stream (left) and a losing stream (right).

The Akaki river may temporarily become a losing stream. When the hydraulic gradient in the aquifer
adjacent to the river is reversed due to draw down of water table during the dry seasons of the year, water
flows from the river into the groundwater. This might also be promoted through increasing pumping rates
in wells (Aynelam, 1999). Surface water pollution has already recorded in many parts of Addis Ababa
(Tamiru, 2001). Groundwater pollution is also becoming a major threat particularly where the groundwater
table and the surface water coincide, like around Kality (Gizaw, 2002). AAWSA et al. (2000) as cited in
Tamiru (2001) have pointed out that the aquifer near Kality feeds both the Little Akaki and Big Akaki
rivers and their tributaries, from their headwaters up to a point near Akaki bridge, as the groundwater level
is higher than the river bed level, along their courses, although they recommended further investigation.
Downstream of this point and up to Aba Samuel hydropower plant, the groundwater level becomes lower
than the river bed. It is believed that there is no hydraulic connection between the aquifer and the surface
water bodies in this part.

However, previous investigations (Aynelam, 1999) and the current study show that there is possibility of
leakage through the deep cutting fractures even down the Akaki bridge. However, the extent of interaction
of the river and the groundwater system may show seasonal variation. The pH, EC, TDS, and total coliform
concentration in the groundwater reflect the influence imposed by polluted surface water, implying the
strong seepage of surface water into the groundwater system .Such mixing theory is corroborated by actual
chemical, e.g., TDS versus ionic concentration, and stable isotope data (Aynelam, 1999). A linear
correlation between conservative constituents further indicates mixing (Gizaw, 2002).
The rate of seepage is often greatest in areas where wave action may restrict the deposition of finer
sediments. Therefore, the locations where the polluted Akaki River interacts with the underlying
groundwater vary from place to place.

Aquifer types (confined leaky and unconfined), and thickness of unsaturated zone have an effect on
movement of contaminations in the subsurface. Even though the unsaturated zone in the wellfield is thick
(30-60 meters), tectonic activity (structures) in the area create favorable pathways for transport.
The thick unsaturated zone in the wellfield can act as a geochemical and bacteriological filter, because of
its low permeability (mostly black cotton soil). The aquiclude (massive rocks or clays) prevents both
downward and upward groundwater flow from the surface to an aquifer and from a deep aquifer to a
shallow aquifer, respectively. It is, therefore, unlikely that the contaminated surface water flows into the
aquifer in areas where these units are found. Though the leakage of contamination can be attenuated by the
black cotton soil in some localities, the Akaki river could still have impact on the surrounding alluvial
aquifer.

87
6.3.3. Geochemical processes affecting transport
The relative abundance of ions in groundwater is determined by the geochemical reactions as among the
groundwater and the various minerals in the aquifer media. Geochemical reactions such as hydrolysis and
complaxation, precipitation/dissolution, oxidation/reduction, sorption, as well as advection and
hydrodynamic dispersion processes all affect the movement of contaminants in the environment. However,
the relatively large number of contaminants in the water does not allow at this stage to identify the
chemical behaviour of each of them. Part of the water in the Akaki wellfield flows by advection and
hydrodynamic dispersion from the aquifer situated under the city of Addis Ababa and Akaki town which
are potential pollution source areas. Therefore, the quality of groundwater located upstream of the wellfield
can have an impact on the quality of water in the wellfield. In this study, only an advection process is dealt
with.

6.3.4. Effect of the slope on transport

Slope variability of the land surface is an important factor in groundwater vulnerability and pollution
assessment as it determines the amount of surface runoff produced, the precipitation rate and displacement
velocity of the contaminant (Civita and De Maio, 2000 cited in Tamiru, et al., 2005). Furthermore, the
slope may be genetic factor of the soil type and thickness that indirectly facilitate the attenuation potential
of the hydrogeological system. Slope also determines the extent of runoff of the pollutant and the degree of
settling sufficient for infiltration. Areas with steep slopes, having large amounts of runoff and smaller
amounts of infiltration, are less vulnerable to groundwater contamination (Napolitano, 1995 cited in
Tamiru, et al., 2005). Generally, low to gentle slopes, i.e., surface zones where a pollutant may be less
displaced under gravity action are highly vulnerable. The study area which is dominated by gentle slopes,
except few steep slopes in the south and east, is highly vulnerable to groundwater contamination.

6.4. Human activities having pollution potential in the area

Factories that dispose untreated effluents and household sewage have been causing pertinent and wide
spread surface water contamination in Addis Ababa. The Kaliti sewer treatment plant, located south of the
city and northeast of the Akaki wellfield, receives waste from sewer lines and waste disposal trucks. Water
used for washing and cleaning, heating and cooling processes, is disposed without treatment from
industries contaminated with various chemicals. All possible sources of contamination including industries
(steel, pulp, paper, pigments, caustic soda paint, pump, brewing, textile, food processing, and meat packing
factories); dairy farms, open-air slaughtering, quarries, agricultural plots, grave yards, dense settlements,
and open market areas are prevalent in the area. The main polluting industries are generally aligned along
the Addis Ababa-Debre Zeit road.
The small agricultural plots are irrigated with either the river water which is contaminated with toxic
substances dumped into it from the close by industries (particularly the Akaki Textile Factory) and/or
through industrial liquid waste directly applied on the farmlands.

The sewage collected using vacuum trucks is discharged into drying beds constructed near the Kaliti waste
stabilization pond. The chemical composition of the river water, therefore, likely represents the mix of
natural as well as artificially induced ions.

88
6.5. Locations of potential contaminations
The following locations are selected as potential contaminant areas that may have an impact on the
wellfield. These are
ƒ Tulu Dimtu scoria which is highly fractured, porous and permeable rock sequence; it is located on a
relatively elevated topography and the beds are partly tilted; the site has been used as a grave site; the
wellfield is found at a lower elevation close to the foot of the hill;
ƒ Gelan metal industry (located at 480653, 976985; elevation, 2130 m.a.s.l) is located on the way to
Debrezeit road near Tulu Dimtu scoria, adjacent to Dengora stream to which it releases its liquid
waste; the Dongora stream then crosses through the center of the wellfield downstream the factory;
ƒ Kality treatment plant to which the highly polluted rivers that drain the Addis Ababa city, most
sewerage lines, and few sanitation systems are directed; and
ƒ Akaki Mesfin Zelelew dairy farm (located at 481507, 976220; elevation 2100 m.a.s.l) to know the
potential leakage of pollutants from the farm (bacteria, animal wastes, etc).

6.6. Predictive simulations of contaminant analysis

Solute transport simulation provides an ideal means to synthesize the controlling processes, evaluate their
interactions, and test the effectiveness of remedial measures. The present study investigates the travel time
of contaminants from their sources to the wellfield, the recharge, discharge and path lines of the
contamination.

Figure 75. Simulation of GW flow & solute transport (Chiang & Kinzelbach (1998)

89
Particles are injected at selected locations and their travel times as well as their path lines are calculated by
running particles forward. Contaminant locations which affect the wellfield are then distinguished from
those which do not. Delineation of capture zones of the pumping wells has been conducted by using
PMPATH which loads the current model automatically, where particles are placed around the pumping
wells. The capture zones of different years are therefore examined by running particles backward.
Since contaminations are mostly considered to be from a surface source, they are placed only on a top cell
face (face 5). See fig. 76.

Figure 76. Various faces of an individual cell.

6.7. Pollutant travel time and direction of tracking computation

PMPATH provides the option of tracking particles forward in the direction of groundwater flow, or
backward toward points of recharge. For backward tracking, particles terminate at points of recharge, rather
than points of discharge. The backward tracking option often provides an efficient means of delineating the
source of recharge to localized points of discharge, such as wellfields or drains.

Contaminants are introduced at sources upstream in cells (40,18), (35,39), (33, 35) and (31, 26) and the
distance of travel of contaminants through the steady state flow field is observed for 100, 150, 800, 300 and
30 years travel times, respectively (Figures 78 - 82). )LJXUH shows that the simulated groundwater
flow direction is to the wellfield almost from all directions and the velocity is relatively higher in areas
where there is a high gradient. In the other areas it has relatively slow velocity as can be seen from the
length of velocity vectors to the extent they seem dots in most areas. Therefore, any contaminants released
at these two velocity locations will have different travel times, with short travel time corresponding to the
high velocity areas and long travel times corresponding to low velocity areas.

90
 
Figure 77. Flow nets showing column and row projections

Kuye Site

Tulu
Dimtu
Scoria

Gelan Site

Mesfin Dairy farm

Figure 78. Contaminant path lines of 100 years travel time, from pollutant sites

91
Figure 78 clarifies the path that contaminant follows at time duration of 100 years from pollutant sites
such as Tulu Dimtu scoria grave yard, cell (31, 26), Mesfin Zelelew dairy farm, Gelan metal factory, cell
(35, 39) and kuye site, cell (40,18). The contaminant path lines from Tilu Dimtu scoria get very close to
one of the wells in the wellfield. (Note that in figure 78 and Figures 79-81 (i) the velocity vectors show the
direction and magnitude of groundwater and contaminant flow velocity. The length of arrows indicates
relative magnitude of groundwater and contaminant velocity. The direction of arrows indicates local flow
direction during the given stress period and time step; (ii) the green lines represent contaminant flow
(stream) lines moving with the same velocity as groundwater, the dark arrows represent velocity vectors
and black line in the cross-sections represents the groundwater surface (potentiometric surface); and (iii)
the cross-section shows the projection of row and column through the center of the wellfield.

Figure 79. Contaminant migration from pollutant sites with in 150 years.

Pollutants start to enter approximately at about 150 years to the wellfield from both Mesfin dairy farm and
Gelan metal industry. But contaminant from Tilu Dimtu scoria has entered into one of the wells in the
wellfield at 100 years time span.

92
 Kuye Site

Figure 80. Contaminant entering to wellfield from kuye site at about 800 years.

Kality
treatment
plant

Figure 81. Pollutant migration from Kality treatment plant for 300 years.

93
The pollutants coming from factories in the Kality area indicates that it does not have much effect to the
wellfield contamination if not they are mixed with the surface water from Akaki catchment which pollutes
wellfield in about 30 years time.

6.8. Implementation of groundwater protection zoning in Akaki wellfield

Groundwater protection zoning is a supplemental methodology for groundwater protection that includes
land use planning in groundwater management. Long-term protection through the use of protection zones
will be balanced by the need for economic and social development by allowing more activities outside the
protection zones. The delineation of a protection zone is the process that determines the geographical area
that should be included in a protection zone program. This area of land is then managed to minimize the
potential of groundwater contamination by human activities that occur on the land surface or in the
subsurface. Proper implementation of aquifer protection zoning will ensure water quality benefits in the
long term. As a result, the unpolluted water sources will aid in good health of the people, animals and
ecosystems. Additional secondary benefits are a healthier workforce and living Environment and can add
significantly to the economic well-being of the area. Effective and focused protection can be achieved
through a differentiated protection approach such as aquifer protection zoning, where the local importance
of aquifers is considered (DWAF, 2000). The communication of the benefits must be to communities,
interested groups and policy makers for them to start implementing the protection of water resources on all
management levels.

Benefits
Proper implementation of aquifer protection zoning will ensure water quality benefits in the long term. As a
result, the unpolluted water sources will aid in good health of the people, animals and ecosystems.
Additional secondary benefits are a healthier workforce and living environment and can add significantly
to the economic benefits. Ecosystem benefits, health benefits to the users and financial savings to the
management institutions are some of the benefits of properly implemented protection zoning. These
benefits enable the recovery of implementation cost and strengthen the urgency and importance of the
implementation of protection zoning. Still, zoning measures need local understanding, acceptance and
control to be able to be meaningful. Appropriate participation of the key stakeholders and the general
public is a requisite for sustainable development of a scarce resource.

Monitoring of protection zone status

An essential component of any groundwater protection programme is water level and –quality monitoring.
This is required to assess the initial conditions and to confirm the effectiveness of the protection measures.
A monitoring strategy for an aquifer protection zone is generally designed to perform three functions -
source release detection, ambient trend monitoring, and early warning detection (Carter et al., 1987).
Verification monitoring is needed to quantify uncertainties in many of the more complex aquifers,
especially fractured aquifers (Muldoon. & Bradbury, 2005; Xu and Van Tonder, 2002).

The monitoring within each aquifer protection zone must include (US-EPA, 2004):
1) Monitoring of the chemical composition of the groundwater, including specification of the contributions
from human activity.
2) Monitoring of the associated surface systems, including terrestrial ecosystems and bodies
of surface water, with which the groundwater body is dynamically linked.

94
3) Monitoring of the water balance components, which should include water levels, rainfall,
discharge and abstraction. These monitoring programmes should be integrated with local protection
measures, as in Water Services Plans, Catchment Management Strategies.

Identify potential threats

An inventory of the potential threats to the catchments long-term sustainability needs to be maintained.
This must include potential point and non-point sources of contamination as well as activities that impact
water quantity, such as urban development. Contamination of groundwater sources has been observed
world-wide, and it is becoming self-evident that concentrated human activity will lead to even more
groundwater contamination. Groundwater studies in several African countries show that the contamination
of water-supply aquifers is mainly due to the improper placement of land-based activities such as
agriculture, industries, waste disposal (Banoeng-Yakubo et al., 2006; Boukari et al., 2006; Nkhuwa, 2006;
Usher et al., 2004) and excreta disposal (Nkhuwa, 2006; Vogel et al., 2006). In Zambia and Ethiopia the
presence of human in the aquifer recharge areas have contaminated groundwater with serious public health
implications and risks for users in the future (Nkhuwa, 2006; Tamiru et al., 2005).
Contamination of drinking water occurs when all three the following components exist;
• A potential source of contamination,
• A pathway to an underlying aquifer and
• A potential user of the water.

Groundwater protection zoning is based on the identification and understandings of groundwater flow
paths from recharge areas towards potential users. The potential threats can be ranked according to their
degree of risk in impairing water sources. Areas in which water sources are vulnerable to these threats can
now be identified. Decision makers can use this information to prioritize and decide which threats need to
be managed most immediately to prevent, reduce or eliminate risks to water sources.

6.9. Capture zone of the wellfield

The capture zone or contributing area of a ground water extraction well can be defined as that portion of
the aquifer from which the well draws its water. Accurate delineation of capture zones is important in many
ground water remediation applications and in the definition of wellhead protection areas. The delineation
of capture zone is often simplified by using steady-state model based on time-weighted average pumping
rates and background hydraulic gradients.

Due to the unique nature of the aquifer type in the area in that it has high ranging between 10 m2/day to
4200 m2/day and the prevalent tectonic features (aligned scoria cones, fractures and faults mainly due to the
effect of Main Ethiopian Rift) in the Akaki wellfield, major contamination is expected to come from areas
outside model domain via Akaki rivers and also from factories nearby. As it can be seen from the figure 82,
groundwater is converging towards wellfield from every direction due to higher rate of pumpage from
wellfield area. This all areas can be considered as the capture zone of groundwater. This all areas should be
protected for safe management of groundwater resource management at the wellfield. Safe capture zone of
wellfield without pulling contaminant towards itself is 30 years. After 30 years it starts to suck contaminant
from contaminant source particularly from polluted big Akaki river. In figure 82, the green lines indicate
that the path lines that contaminant can follow with in 30 years span. This is done by putting particle at the

95
wells in the wellfield and tracking particles backward to see at which time duration contaminants from their
source will reach the wells in the wellfield.

Figure 82. Contamination capture (green) lines of wellfield for 30 years.

96
 
Figure 83. Capture zone of wellfield for 300 years

In this case contaminants from most of the study areas especially from polluted Akaki river and factorie
around the wellfield will be captured in the wells. In contrary to this, within 30 years, only areas close to
the wellfield are captured.

Protection zone delineation

Many countries world-wide have implemented borehole protection zones, also called groundwater supply
protection areas or wellhead protection zones, with the special focus of protecting domestic water supplies
against pollution (Foster et al., 2002).
These borehole protection zones have to protect the groundwater from:
• contaminants that decay with time, where subsurface residence time is the best measure of
protection; and
• non-degradable contaminants, where flow path-dependent dilution must be provided.

In Ethiopia, particularly in Akaki wellfield aquifers, until today there is no strict zone which is delineated
to protect water supply wells against pollution.

97
6.10. Delineation of a 30 year protection zone of Akaki wellfield
The steady state model is used to delineate a 30 year protection zone, for Akaki wellfield. PMPATH
program is employed to simulate path lines by placing particles at borehole sites and tracking them
backward over a period of 30 years. The basic assumption regarding this procedure is that the contaminant
(example NO3-) moves at the velocity of groundwater flow. In addition, major changes in abstraction rates
are assumed. This is because great changes in abstraction rates may greatly affect the average seepage
velocity of the groundwater, thereby resulting in a different flow pattern and hence the resulting path lines.
Figure 84. Shows a 30 year protection zone for boreholes in the wellfield with which a potential source of
contaminant should not be placed. The spatial distribution of travel time through the aquifer was obtained
by placing particles in appropriate model cells, and tracking their movement backward through time to their
point of origin. Generally, the smaller the travel time in a location, the more susceptible is the aquifer to
potential contamination at that point.

30 years safe
capture zone

Figure 84. A 30 year protection zone for boreholes in the wellfield

The maximum years that the wellfield can be safe from entering of contaminants from any sources
surrounding it with current abstraction rates is 30 years. For more than 30 years duration, contaminants will
try to enter in to some of the wells from polluted Akaki river. From other sources of pollution will not
pollute for approximately 100 years.


98


Figure 85. Protection zones delineated around the wellfield.

The protection zones as can be seen in the figure 85, which are represented as zones “A” and “B” should be
strictly protected from any pollution sources to the groundwater. Any factories or any pollutant sources
should not be constructed in this zone. The upstream of Akaki river should be kept clean to prevent its
pollution which can in turn contaminate the wellfield which is operating at pumping rate of more than
46000 m3/day.


99
7. Conclusions, discussions and
recommendations

7.1. Discussion and conclusions

The Akaki river catchment, which includes the city of Addis Ababa and the Akaki wellfield area, has an
extensive drainage system which covers an area of 1500 km2 with an elevation drop of over 1000 m over a
distance of about 30 km.

As Addis Ababa is situated at the western shoulder of the main Ethiopian rift, the rocks are strongly
affected by the rift tectonics, which is manifested by the number of faults following the general trend of the
rift system (NE-SW). The city has a relief of rather flat areas, alternating with volcanic hills and
surrounded by trachytic mountains. In general, the Addis Ababa area is underlain by a complex system of
aquifers of various size and hydraulic properties, which are mostly hydraulically connected. The aquifer of
Akaki wellfield is rather assumed to be made up of many pyroclastic deposits such as scoria, fractured
scoraceous and vesicular basalts, ignimbrites, tuff and ashes (AAWSA, 2000).

The annual rainfall values in the catchment are observed to increase with elevation with correlation
coefficient of 0.96. Therefore, the hypsometric method is used for rainfall estimation in the catchment.
The mean annual rainfall determined by this method is 1224 mm/yr with maximum and minimum RF of
1705 and 970 mm/yr.

The hydrochemical characteristics of groundwater in Akaki catchment volcanic aquifer reflect the
interaction of various natural processes and human activities. The overall hydrochemical evolution with in
the study catchment is still at early stage, however, because of variations in length of flow path, flow
velocity and residence time, a clear evolution from north to south is observed. Less chemically evolved
groundwater is represented by a predominant Ca-Na-HCO3 chemical water type dilute system draining
from mountainous region of the Intoto silicics, marking typically the initial recharge area. With increased
residence times and longer flow paths, groundwater interacts with more aquifer materials and acquires
more chemical species with higher concentrations. This is highly reflected by groundwater draining from
the young scoria, scoraceous and vesicular basalt located at the south southern part of the catchment (Akaki
and Kaliti areas) particularly around the wellfield. Water type around these areas reflects mixed cation
bicarbonates with more calcium and magnesium cations indicating much longer residence time and reaction
of water with aquifer system. River water samples taken from Akaki bridge reveal Mg-NO3-Cl and other
rivers in the area show Ca-Mg-Na-HCO3 and Ca-Na-HCO3 type. A shift of water chemistry towards SO4+2
Cl and NO3-1 is mostly related to the infiltration of contaminants into the rivers or subsurface rather than
natural dissolution processes.

The aquifer system is multilayered, but since almost all of the boreholes are screened at more than one
aquifer layer, the aquifer parameters reported represent the screened layers all together. Therefore,
treatment of each aquifer layer is practically impossible at this stage. Because of this, modeling at the
Akaki catchment and wellfield is done taking the screened aquifers as single layer. Well logs are used to

100
assist in the conceptualization of the system prior to accomplish modeling works in wellfield and its
catchment.

The area is made up of multi-aquifers having different hydrogeological characteristics and all types of
aquifers are found in the area (confined, semic-confined and confined) including perched aquifers.

The Chloride Mass Balance Method (CMB), semi-distributed water balance model and recursive digital
filter methods were employed to estimate the recharge in the catchment.
Mean annual groundwater recharge based on CMB is 326 mm year-1 which is 24 % of the average annual
rainfall in the area. This figure sets the upper limit to the recharge because effluents from the city are also
infiltrating to groundwater elevating the chloride contents. Chloride mass balance assumes that all the
rainfall in the catchment infiltrates and contributes to recharge. It also assumes that no additional source of
chloride is available besides rain chloride. However, these two assumptions are not valid in this catchment
because of a total runoff of about 492 MCM/yr which results from direct runoff and urban water return
flow. The city of Addis Ababa is located in the catchment and it has unregulated solid and liquid waste
disposal in the catchment leading in elevated chloride concentrations in both groundwater and surface
water sources which also falsifies another assumption in to use CMB. As the result of this, recharge is
overestimated by this method for this particular catchment. To integrate with CMB, another two methods
are adapted in present study to evaluate recharge namely: semi-distributed water balance model and
recursive digital filter methods. From the water balance as simulated by the catchment’s model, the total
average base flow estimated is 92 MCM/yr or 62 mm/yr. From the water balance model described in
chapter 3, a base flow and recharge value of 130 MCM/yr or 87 mm/yr is found. Using the digital filter
method, the base flow and recharge values are roughly 65 MCM/yr (63mm/yr) indicating good agreement
between the models.

The aquifer parameters obtained from the pumping test result show very high contrasting hydraulic
properties spatially. There are extremely high and low values of transmissivity and hydraulic conductivity
values obtained from nearby wells. The assumptions of homogeneity and infinite horizontal extent of
aquifer usually considered in pumping test data analysis are not valid here due the geological heterogeneity
of the area. Thus it is tried to optimize the hydraulic properties during the calibration process using the
calibration target (hydraulic heads).
The water budget in the catchment model agrees with the modeling results obtained from other models as
explained in chapters 3, 4 and 5.
Concerning the modelling works in this study the following issues should be pointed out:
1. The Akaki wellfield is highly vulnerable to contamination from surface waters and direct
infiltration. Even though the groundwater level around the Akaki River is located at about 30 m
below the river bed, there is a hydraulic link between the river and groundwater through rock
fractures. The river has direct impact on the nearby wells that tap water from alluvial layer. The
quality of the surface and groundwater up stream will determine the quality of water in the
wellfield.
2. In the model domain, the groundwater flow lines converge towards Akaki wellfield from all
directions, implying that contaminants are carried into the wellfield from all directions.
3. The intensive pumping of groundwater from the Akaki wellfield results in rapid decline of
groundwater levels, leading to disturbance in the steady state flow system of the groundwater,
eventually resulting in increased velocity of groundwater flow towards the depression zone. This
101
 process potentially facilitates the rapid flow of contaminated water from upstream sections of the
aquifer to the wellfield.
4. Currently, the degree of contamination of the groundwater is negligible giving certain time to
avert the problem. However, the model indicates that the groundwater path lines with
contaminated water injected at contaminant sources upstream will reach in the wellfield in less
than 30 years time.
5. Flow velocities are different throughout the system. They are more rapid in areas where porosity
and transmissivity are high (wellfield and Akaki town) and lower in relatively flat areas (Fanta
and Kuye areas).
6. Generally, though urbanization and industrialization are the immediate sources of contamination,
the geological and hydrogeological character of the area facilitates contaminant migration.

7.2. Recommendations
Several assumptions had to be made for the hydrological modelling purpose in the present study because of
lack of sufficient knowledge about interactions between different hydrological processes. Data inadequacy
was also encountered critically. Consequently, further works need to be undertaken in data collection and
field investigations, among which the following are worth mentioning:

The gauging station at Aba Samuel has to be calibrated using a wide range of stage and discharge data, as
this gauging station is highly important for the study of the Akaki groundwater as well as stream-aquifer
interactions. The rating curves of gauging stations on Mutinicha river at the confluence point with Akaki
River downstream of Legedadi dam and Little Akaki river at Asco Tunnery should be checked regularly
and recalibrated if necessary so that they can produce good quality data over a wide range of stage
discharge values.

Recharge mechanisms of Akaki aquifer should be investigated further. This requires a detailed
understanding of the hydrogeology of the region. The spatial variation of permeability should also be
studied in depth through field and laboratory tests.
Knowledge about stream-aquifer interaction, particularly in the wellfield area, should develop further in
order to improve the representation of the actual aquifer by the conceptual model and assess the risk of
pollution of the aquifer by Akaki river more accurately. The piezometers planned to be drilled in the
wellfield area with the assistance of International Atomic Energy Agency (IAEA) would be of great value
in this regard. Integrated surface water – groundwater-modelling approach is expected to give a better
representation of the reality. Therefore, attempt should be made to refine the groundwater modelling
procedure in that direction.

To model the catchment as well as wellfield with reduced model uncertainties, the following should be
recomended first.
ƒ Further work is required to define the thickness of each formation because it has significance
in the distribution of hydrogeological properties.
ƒ The clear demarcation between different formations requires further detail study.
ƒ The aquifer geometry and layers need detailed study and investigation.
ƒ Thoroughly understanding of the multi- layer volcanic aquifer in the area, this may be used in
the assessment of the extent to which pollutants percolate in the subsurface.

102
 ƒ Further chemical and isotopic analysis has to be adopted for groundwater flow paths and
hydrodynamics analysis.
ƒ Application of mass-balance and flow simulation (mixing model) calculations to
interpretations of mixing between different water types including surface and ground waters.
ƒ Continuous monitoring of water quality and water level must be conducted so as to determine
the groundwater dynamics.
In view of tackling the actual and potential contamination risks of the Akaki wellfield and the Akaki river
catchment in general, more effort should be made to address the following points:
ƒ Unless proper groundwater management strategy is implemented, the groundwater resource
may become at risk. Presently there is no controlling and management mechanism for the
aquifer in the studied area. Therefore, it would be important at this juncture to implement
appropriate management and controlling mechanism for the groundwater resource.
ƒ Conduct contaminant transport analysis taking into account chemical reactions, attenuation and
multiple layer aquifer structures;
ƒ Create closed–loop water supply systems at industrial enterprises involving effluent reuse;
ƒ 'HVLJQ DQG LPSURYH VHZHUDJH DQG ODQGILOO V\VWHPV LQ $GGLV $EDED FLW\ DQG
$NDNLWRZQUHGXFLQJWKHDPRXQWRISROOXWLRQRYHUWLPH
ƒ Implement strict environmental policy (e.g., type of fertilizers, industries, agricultural practices
etc. to be allowed in the area) on the pumping sites with particular emphasis on the protection
zones established around the wellfield;
ƒ Limit activities having pollution potential in special areas sufficiently far from water supply
wells;
ƒ Introduce and implement legal control mechanisms (e.g., strict rules on treatment of effluents
before disposal);
ƒ Set standards to discharges of potential bodies and penalties for non-observance of
requirements; and
ƒ Closely monitor the chemical quality of groundwater in the wellfield and surrounding areas in
addition to the surface waters.


103
 References
AAWSA and SEURECA, 1989. Addis Ababa Water Supply Project Stage IIIA, Feasibility Study and
Preliminary Design, Vol. 7, Addis Ababa.
AAWSA-SEURECA, 1991. Addis Ababa Water Supply Project Stage III, Feasibility Study and
Preliminary Design. Groundwater Resources. V.IV. Addis Ababa. 135pp.
AAWSA, 2000. Addis Ababa water supply project Stage-IIIA Groundwater-Phase II, main report, Addis
Ababa Water and Sewerage Authority, Addis Ababa, Ethiopia, pp 67.
Anderson, M.P and Woessner, W.W., 1992. Applied groundwater modelling: Simulation of flow and
advective transport. San Diego Academic Press, 381 pp
Aquaterra Consulting Pty Ltd., 2000. Groundwater flow modelling guideline, Western Australia.
http://www.mdbc.gov.au/_data/page/127/model_guide.pdf
AG Consult, 2004. Hydrogeological map of Addis Ababa, Akaki and Dukem areas. AG Consult, Addis
Ababa.
Anteneh, G., 1994. Hydrogeology of Akaki Area, MSc. Thesis, Addis Ababa University, Addis Ababa, pp
139.
Aynalem, A., 1999. Water Quality and Ground Water/Akaki River Interaction in the Sekelo Basin (Lower
Akaki River Sub Basin), MSc. Thesis, Addis Ababa University, Addis Ababa, . pp 116.
Banoeng-Yakubo, B.K., Akabzaa, M., Hotor, V. & Danso, S.D., 2006. Application of electrical resistivity
techniques in delineation of saltwater-freshwater in Keta Basin, Ghana. Xu, Y. & Usher, B. (ed),
Groundwater Pollution in Africa. London: Taylor & Francis/Balkema.
Berhane Melake, 1982: Hydrogeology of Upper Awash Basin Upstream of Koka Dam. Ministry of Mines
and Energy. pp. 171.
Carter, L.W., Knox, R.C., and Fairchild, D.M., 1987. Ground Water Quality Protection. Lewis Publishers,
Inc.; Chelsea, Michigan. 120 pp.
Chiang W.H. and W. Kinzelbach, W., 1998. Processing Modflow a simulation system for modeling
groundwater flow and pollution. Software manual: 325
Civita and DeMario, 1998. Mapping Groundwater Vulnerability By the Point Count System. Italy.
Daniel, G., 1977. Aspects of Climate and water budget in Ethiopia. A technical Monograph Published for
Addis Ababa University. Addis Ababa University Press, 127 pp.
Das, D. B., and Lewis, M., 2007. Dynamics of fluid circulation in coupled free and heterogeneous porous
domains. Chemical Engineering Science, 62(13), 3549-3573.
Davis, S.N and Dewiest, R.J.M., 1966. Hydrogeology. John Wiley and Sons, Inc. New York, 463pp
Deckers, D. L., 2006. Predicting discharge at ungauged catchments. Civil Engineering and Management
Enschede, University of Twente. MSc Thesis.
Deme, I., Tandia, A.A., Faye, A., Malou, R., Dia, I., Diallo, M.S. and Sarr, M., 2006. Management of
nitrate pollution of groundwater in African cities: The case of Dakar, Senegal. Xu, Y. and Usher,
B. (ed), Groundwater Pollution in Africa. London: Taylor and Francis/Balkema.
Demlie, M., 2007. Hydrogeology of Akaki catchment, central Ethiopia: with special emphasis on the
problem of groundwater recharge, PhD Thesis, Ruhr-University, Bochumer, Germany, 153p.
Dereje Nigussie, 2003. GIS based groundwater vulnerability assessment in the Akaki river catchment
(Addis Ababa), central Ethiopia. M.Sc. Thesis, school of graduate studies Addis Ababa University,
Addis Ababa, 163 pp.

DWAF, 2000. Policy and Strategy for Groundwater Quality Management in South Africa. Water
Quality Management series. Department of Water Affairs and Forestry, Pretoria. South Africa.

104
EPA, 1997. Environmental policy of Ethiopia, Addis Ababa, Ethiopia. 198 pp.
Edmunds, W.M. and Gaye, C.B., 1994. Estimating spatial variability of groundwater recharge in the Sahel
using chloride. Journal of Hydrology, 156:47-59.
Eriksson, E. (1985). Principles and applications of hydrochemistry. Chapman & Hall, London-New York.
Harbaugh AW, Banta ER, Hill MC and McDonald, MG., 2000. MODFLOW-2000, The U.S. Geological
Survey modular ground-water model User guide to modularization concepts and the ground-water
flow process, U. S. Geological Survey, Open-file report 00-92, pp 410.
Http/upload wikimedia.org/wikipedia/en/thumb/d/d1/Effective porosity 2jpg/800px-Effective
porosity 2 jpg, 5/29/2007.
EMA, 1988: National Atlas of Ethiopia. Ethiopian mapping Authority, Addis Ababa.
Eccleston, B., 1997. Water Supply for Kaliti cotton mill, Akaki, near Addis Ababa, Ethiopia, 20p.
Eckhardt, K., 2005. How to construct recursive digital filters for baseflow separation. Hydrological
Processes,V. 19 (2): 507-515.
Edmunds, W. M., 1981. Hydrogeochemical investigations: case studies in groundwater resource
evaluation, Lloyd, J.W. (edit). Oxford University press, Oxford, 87-112pp.
Eriksson, E., 1985. Principles and applications of hydrochemistry. Chapman and Hall, London-New York.
Fetter, C. W., 2001. Applied hydrogeology: Fourth edition. Prentis-Hall, Inc. Upper Saddle River, New
Jersey, 598pp.
Freeze, R. and Cherry A., 1979. Groundwater. A Simon and Schuster Company Englewood Cliffs, USA.
604pp.
Foster, S., Hirata, R., Games, D., D’Elia, M. and Paris, M. (2002). Groundwater Quality Protection – A
guide for water utilities, Municipal on authorities, and environment agencies. Groundwater
Management Advisory Team (GW. Mate. Washington: The World Bank).
Getachew Asmare, 2005. Model Based Groundwater System Analysis for Hayk-Ardibo Catchment, MSc.
Thesis, Addis Ababa University, Addis Ababa, pp 104.
Gebrekidan, H., 2000. Investigation of the Ionic Composition of Addis Ababa drinking and Surface Waters,
Msc. Dissertation. 60p.
Gieske, A.S.M., 1992. “Dynamic of groundwater recharge” case study in Semi arid, Botswana. PhD Thesis,
Free University of Amsterdam, Amsterdam, The Netherlands. 289 pp.
Gizaw, B., 2002. Hydrochemical and Environmental Investigation of the Addis Ababa Region, Ethiopia,
PhD Thesis, Faculty of Earth and Environmental Sciences, Ludwig-Maximilians-University of
Munich, Germany. pp 188.
Girma, A., 1994. Hydrogeology of Akaki Area, MSc. Thesis, Addis Ababa University, Addis Ababa.
136 pp.
Girmay, H. and Assefa, G., 1989. The Addis Ababa-Nazareth Volcanics: A Miocene-Pleistocene Volcanic
Succession in Ethiopian Rift. SINET: Ethiopian Journal of Science 12 (1): 1- 24.
Hounslow, A. W., 1995. Water quality data: analysis and interpretation. Boca Raton etc.: CRC Lewis.
ITC, 2001. ILWIS 3.0 Academic: User's guide, Enschede, The Netherlands.
Kazmin, V., 1975. Explanation of geological map of Ethiopia.Geological Survey of Ethiopia. 120 pp.
Lerner, D.N.,A.S.Isa and Simmers I., 1990. Groundwater recharge: a guide in understanding and estimating
natural recharge. International association of hydrogeologists, Hanover, Germany.
Lyne, V and Hollick M., 1979. Stochastic time-variable rainfall-runoff. Institute of Engineers Australia
National Conference. Publ. 79/10, 89-93.

105
Lubczynski, M., 1997. Application of numerical flow modelling combined with remote sensing and GIS
techniques for the quantification of regional groundwater resources in hard rock terrains, pp. 151-
157.
Lundmark, K.W., Pohll., G.M. and Carroll, R.W.H.,2007. A steady-state water budget accounting model
for carbonate aquifer system in white pin country, Nevada and adjucent areas in Nevada and Utah.
Logan, J., 1964. Estimating transmisivity from routine production wells. Groundwater 2(1): 35-37.
McDonald, M. G., and Harbaugh, A. W., 1988. A modular three-dimensional finite difference ground-
water flow model, U.S., Geological Survey.
Mohr, P.A., 1983. The Ethiopian Flood Basalt Province, Nature. Vol., 303: pp 577-584.
Morton W.H., R.D.C., Mitchelljg., and Mohr P.A., 1966: “Rift Ward Younging of Volcanic Rocks in the
Addis Ababa Region, Ethiopian Rift Valley”,134 pp.
Muldoon, M. and Bradbury, K., 2005. Site Characterization in Densely Fractured Dolomite: Comparison of
Methods. Ground Water Vol. 43, No. 6, November–December 2005. National Ground Water
Association.
Nkhuwa, D.C.W., 2005. Groundwater quality assessments in the John Laing and Misisi areas of Lusaka.
Xu, Y. and Usher, B. (ed), Groundwater Pollution in Africa. London: Taylor and
Francis/Balkema.
Obakgeng, O.T., 2000: Groundwater recharge and vulnerability: a case study at the margins of the south
east centeral Kalahari sub basin, Serowe region, Botswana. MSc. Thesis, ITC, Enschede, 125pp.
Press et al., 1986: W.H. Press, B.P.Flannery, S.A.Teukolsky and W.T.Vettering, Numerical recipes: the art
of scientific computing, Cambridge Univeristy press, Cambridge, New York.
Rientjes, T.H.M., 2007: in Hydrology: Lecturing materials. ITC, Enschede.
Rushton, K.R., Eilers, V.H.M. and Carter, R.C., 2006: Improved soil moisture balance methodology for
recharge estimation. Journal of Hydrology, 318(1-4): 379-399.
Tenalem A., Tamiru A., Yirga T, Solomon W., and Nuri M., 2005: Water quality assessment and
groundwater Vulnerability of Addis Ababa water supply aquifers. UNEP report.
Tamiru A., 2001: The impact of uncontrolled waste disposal on surface water quality in Addis Ababa,
Ethiopia. Ethiopian Journal of Science, 24 (1): 93-104.
Tamiru A., Legesse, D., Ayenew, T., Mohammed, N. and Waltenigus, S., 2005. Degree of groundwater
vulnerability to pollution in Addis Ababa, Ethiopia. pp 111.
US-EPA, 2004. Monitoring Technologies for Wellhead Protection. US Environmental Protection Agency,
Environmental Sciences Division, CMB. Las Vegas, Nevada. 101 pp.
Usher, B.H., Pretorius, J.A., Dennis, I., Jovanovic, N., Clarke, S., Cave, L., Titus, R. and Xu, Y., 2004.
Identification and prioritisation of groundwater contaminants and sources in South Africa's urban
catchment. WRC report 1326/1/04.
Repsold, 1989. Well logging in groundwater development. International association of hydrogeologists,
p.136
Simmers, I., 1988. Estimation of natural groundwater recharge. Dordrecht . D. Reidel.
Simmers, I., Hendrickx, J. M. H., Kruseman, G. P. and Rushton, K. R., 1997. Recharge of phreatic
aquifers in sem -arid areas. International contributions to Hydrogeology: IAH, International
Association of Hydrogeologists; 19. Rotterdam, Balkema etc., 277 pp.
Vernier, A. 1993. Aspects of Ethiopian Hydrogeology : from Geology and mineral resources of Somalia
and its surrounding regions, 1st.Agron.Oltremare, Firenze,Relaze.e, Monogr. 113, 687-698.

106
Vogel, H. Keipeile, K., Kgomanyane, J., Zwikula, T., Pontsho, M., Mafa, B., Matthes,L., Staudt, M. Beger,
K. & Guth, T., 2006. Groundwater Quality Case Studies in Botswana. Xu, Y. and Usher, B. (ed),
Groundwater Pollution in Africa. London: Taylor & Francis/Balkema.
Wood, W.W. and Sanford, W.E., 1995. Chemical and Isotopic methods for quantifying groundwater
recharge in regional semiarid environment. Groundwater.33 (3), 458-468.
Xu, Y & Van Tonder, G.J., 2002. Capture zone simulation for boreholes located in fractured dykes using
the linesink concept, ISSN 0378-4738, Water SA, Vol. 28 No. 2 April 2002.
Yirga Tadesse, 2004. Groundwater modelling a Case study on volcanic water supply aquifer /Akaki well
field/ of The city of Addis Ababa, Ethiopia, pp 21.
Yongxin, X., and Beekman, H. E., 2003. Groundwater recharge estimation in southern Africa. Paris:
United Nations Educational Scientific and Cultural Organization (UNESCO), Paris, 207 pp.
Zanettin B. and Justin Visitin E., 1974. The Volcanic Succession in central Ethiopia: the volcanics of
western Afar and Ethiopian rift margins. Mem. 1st. Geol. Miner. univ. Padova, Italy, V.31:1-19.

107
APPENDICES
Appendix 1 Geologic logs, Resistivity logs, and their corresponding values

:$7(5,,,%25(+2/(%+$
:(//,'(17,),&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB7<3(2)5,*BBBBBBBBBBB
63&+

$.$.,:(//),(/'$5($'
/2&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB67$7,&:$7(5/(9(/BBBBBBBB
P ZDWHUOHY HOGXULQJGULOOLQJ

$$:6$ ; 
2:1(5BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB:(//&225',1$7(BBBBBBBBBB
&203/$17
&2175$&725BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
<  $
%+BBBBB
= P
'$7('5,//('BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
(&
6&5((1'(37+ 0
6&5((16,=( LQFK
'5,//,1*6,=(
&$6,1*6,=(
'(37+P

'(37+P

(/(&75,&
LQFK

352),/( /,7+2/2*<
LQFK

/2**,1*&859(



([SDQVLYHVRLO



 :HDWKHUHGURFN


 $SSDUHQW5HVLVWLYW\
9 9 9 99
9      
9 9 P
 9 9 9 0DVVLYH%DVDOW
9 9
9
9 9
9

 9
3DOHRVRO

 
:HDWKHUHGURFN
)LUVW$TXLIHU



3DOHRVRO




9
9 9


9 9
9
 0DVVLYH%DVDOW
9 9
 :HDWKHUHG%DVDOW
6HFRQGDTXLIHU
 9

9
9 0DVVLYH%DVDOW

 9


3DOHRVRO
 
9
9
0DVVLYH%DVDOW
9 9
 
3DOHRVRO

 0DVVLYH%DVDOW
 9 9
9

3DOHRVRO

 0DVVLYH%DVDOW
9
9 9
 

108
 :$7(5,,,%25(+2/(%+
:(//,'(17,),&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB7<3(2)5,*BBBBBBBBBBB
63&+

$.$.,:(//),(/'$5($'
/2&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB67$7,&:$7(5/(9(/BBBBBBBB
P
$$:6$ ; 
2:1(5BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB:(//&225',1$7(BBBBBBBBBB
&203/$17
&2175$&725BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
(&
< 
= P

%+BBBBB
'$7('5,//('BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
6&5((1'(37+ 0
6&5((16,=( LQFK
'5,//,1*6,=(
&$6,1*6,=(
'(37+P

'(37+P
LQFK

352),/( /,7+2/2*< /,7+2/2*<

LQFK

(/(&75,&
/2**,1*&859(


([SDQVLYHVRLO







:HDWKHUHGURFN



9 9
 0DVVLYH%DVDOW
9 9
9 9
 $SSDUHQW5HVLVWLYW\

3DOHRVRO


P



:HDWKHUHGURFN
  )LUVW$TXLIHU


9 9
9
  0DVVLYH%DVDOW





:HDWKHUHGURFN
 6HF RQG$TXLIHU

 

 3DOHRVRO

9 9
 9
0DVVLYH%DVDOW
9
9

 9 9

:HDWKHUHGURFN


 7KLU G$TXLIHU


 
9 9


9
9 0DVVLYH%DVDOW
9 9

 3DOHRVRO

:HDWKHUHGURFN
  )R XU $ TXLIHU










109
 :$7(5,,,%25(+2/(%+
:(//,'(17,),&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB7<3(2)5,*BBBBBBBBBBB
63&+

$.$.,:(//),(/'$5($'
/2&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB67$7,&:$7(5/(9(/BBBBBBBB
P
$$:6$ ; 
2:1(5BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB:(//&225',1$7(BBBBBBBBBB
&203/$17
&2175$&725BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
< 

%+BBBBB
'$7('5,//('BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
(& = P
6&5((1'(37+ 0
6&5((16,=( LQFK
'5,//,1*6,=(
&$6,1*6,=(
'(37+P

'(37+P

(/(&75,&
LQFK

352),/( /,7+2/2*<
LQFK

/2**,1*&859(


 ([SDQVLYHVRLO




 :HDWKHUHGURFN

9 9 0DVVLYH%DVDOW

9 9 :HDWKHUHGURFN


9 9
 0DVVLYH%DVDOW
9


 :HDWKHUHGURFN

 $SSDUHQW5HVLVWLYW\
9
9 0DVVLYH%DVDOW      
9 P
 



 :HDWKHUHGURFN
)LUVW$TXLIHU

 




9 9 9
 0DVVLYH%DVDOW
  9 9 9

:HDWKHUHGURFN
6HFRQG$TXLIHU
 
9 9 9
0DVVLYH%DVDOW

 :HDWKHUHGURFN
   7KLUG$TXLIHU

9 0DVVLYH%DVDOW
 9 9




 :HDWKHUHGURFN
)RXUWK$TXLIHU





9 9 9
   
0DVVLYH%DVDOW
9 9 9








110
 :$7(5,,,%25(+2/(%+
:(//,'(17,),&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB7<3(2)5,*BBBBBBBBBBB
63&+

$.$.,:(//),(/'$5($'
/2&$7,21BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB67$7,&:$7(5/(9(/BBBBBBBB
P
$$:6$ ; 
2:1(5BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB:(//&225',1$7(BBBBBBBBBB
&203/$17
&2175$&725BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
<  
%+BBBBB
= P
'$7('5,//('BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
(&
6&5 ((1 '(37+ 0
6&5 ((16,=(LQFK
' 5,//,1* 6,=(
&$6,1* 6,=(
' (37+P

' (37+P

(/(&75,&
  LQFK
 LQFK

352 ),/( /,7+2/2*<

/2**,1*&859(

  

([SDQ VLYHVRLO



 

:HDWKHU HG URFN

 

9 9
9
9 0DVVLYH%DVDOW
9

9 9 9
 

 :HDWKHU HG URFN

 
9 9 $SSDUHQ W5 HVLVWL YW\
0DVVLYH%DVDOW
9      
  P
 :HDWKHU HG URFN



)L UV W $TXL IHU
 
9 9 9 0DVVLYH%DVDOW
  
 
3DO HRVRO


  
:HDWKHU HG URFN
  6HFRQ G $TXLIHU
 

9
9 9
99
 9 0DVVLYH %DVDOW
9

  9
9 9
  9
   3DO HRVRO
 
:HDWKHU HG URFN
7K LUG $TXL IHU
 
 9 9 0DVVLYH%DVDOW



9
 


:HDWKHU HG URFN

 ) RXUWK  $TXLIHU

 


 









111
Appendix 2 Electric logging record (resistivity & self potential logging)
BH name: BH5a Depth of Borehole: 168.09m Date:17/11/1997
Measurement Method: Apparent Resistively (No.2M 1.9A) Self potential Logging Vsp (mv)
Depth (m) Vsp (mv) Ps (Ƿm ) Depth (m) Vsp (mv) Ps (Ƿm ) Depth (m) Vsp (mv) Ps (Ƿm )
168 28.2 1042.0 124.0 41.4 69.1 81.0 64.1 360.0
167 26.0 939.5 123.0 43.8 95.2 80.0 53.1 377.0
166 25.7 638.2 122.0 41.6 56.8 79.0 54.6 437.0
165 24.8 36.7 121.0 43.6 418.0 78.0 54.3 319.0
164 23.8 36.3 120.0 87.4 2500.0 77.0 55.1 270.0
163 22.5 36.4 119.0 57.3 1071.0 76.0 55.5 231.0
162 22.3 37.0 118.0 47.6 691.0 75.0 56.8 201.0
161 23.2 45.2 117.0 48.7 733.0 74.0 56.3 1745.0
160 21.3 49.6 116.0 45.4 995.0 73.0 57.7 1237.0
159 23.6 36.8 115.0 45.4 2970.0 72.0 57.2 89.0
158 24.5 24.8 114.0 49.3 2760.0 71.0 58.4 19.2
157 27.2 19.7 113.0 46.8 2440.0 70.0 58.4 22.0
156 24.6 14.8 112.0 50.3 2240.0 69.0 50.7 22.8
155 21.4 20.8 111.0 49.5 1551.0 68.0 60.5 19.7
154 27.4 16.2 110.0 48.5 304.0 67.0 59.8 23.5
153 23.2 35.3 109.0 47.7 1552.0 66.0 60.2 167.3
152 198.9 366.0 108.0 544.0 36.1 65.0 62.6 53.9
151 196.5 1714.0 107.0 47.2 93.0 64.0 61.3 87.8
150 198.8 1461.0 106.0 53.6 66.5 63.0 62.7 42.9
149 197.9 565.0 105.0 53.6 123.8 62.0 60.1 50.4
148 194.7 73.3 104.0 54.4 5150.0 61.0 63.1 27.2
147 25.8 63.2 103.0 47.3 5840.0 60.0 61.3 1511.0
146 25.3 30.0 102.0 53.6 5300.0 59.0 61.0 1436.0
145 22.5 27.0 101.0 56.3 3220.0 58.0 60.9 1626.0
144 26.2 24.0 100.0 56.0 2260.0 57.0 60.5 1627.0
143 25.4 67.5 99.0 56.0 2410.0 56.0 59.7 20.7
142 24.7 60.2 98.0 59.4 3060.0 55.0 61.4 19.6
141 26.9 60.7 97.0 59.1 3430.0 54.0 61.8 31.3
140 30.9 82.7 96.0 58.5 2420.0 53.0 61.8 43.3
139 27.4 104.4 95.0 61.2 2230.0 52.0 60.9 504.0
138 28.5 1221.0 94.0 58.6 2530.0 51.0 59.0 1106.0
137 29.2 1456.0 93.0 58.4 2570.0 50.0 60.2 818.0
136 27.5 847.0 92.0 59.2 2860.0 49.0 63.3 1043.0
135 29.3 1096.0 91.0 60.9 2200.0 48.0 63.2 1025.0
134 30.5 1194.0 90.0 58.9 1732.0 47.0 58.0 244.0
133 34.0 614.0 89.0 60.0 1580.0 46.0 57.3 2130.0
132 30.9 249.0 88.0 63.0 813.0 45.0 55.1 1323.0
131 36.3 180.0 87.0 61.1 632.0 44.0 56.0 1264.0
130 34.9 90.1 86.0 61.1 409.0 43.0 55.3 571.0
129 36.6 56.1 85.0 63.1 1162.0 42.0 55.8 447.0
128 38.5 85.8 84.0 63.9 1055.0 41.0 54.7 1353.0
127 37.6 71.6 83.0 62.5 241.0 40.0 53.2 1647.0
126 38.3 99.3 82.0 62.6 337.0
125 40.1 66.6

112
BH Name: BH 17 Depth of Borehole: 141m Date: 26/10/1996
Measurement Method: Apparent Resistively (No.2M 1.9A) Self potential Logging Vsp (mv)
Depth (m) Vsp (mv) Ps (Ƿm ) Depth (m) Vsp (mv) Ps Depth (m) Vsp (mv) Ps (Ƿm )
(Ƿm )
141 52.7 30.0 111.0 49.8 366.0 81.0 41.1 29.2
140 55.3 30.4 110.0 54.2 2250.0 80.0 29.0 113.7
139 55.1 29.9 109.0 50.6 2150.0 79.0 42.0 108.0
138 57.2 30.7 108.0 47.7 1899.0 78.0 40.2 101.0
137 56.9 24.8 107.0 49.3 1837.0 77.0 41.0 96.2
136 57.1 78.3 106.0 49.0 1605.0 76.0 40.9 92.7
135 56.5 22.0 105.0 47.3 1349.0 75.0 40.4 89.7
134 57.5 86.1 104.0 47.3 1139.0 74.0 39.2 68.4
133 55.4 30.2 103.0 47.1 1057.0 73.0 40.4 55.7
132 53.6 33.1 102.0 46.9 1025.0 72.0 45.6 48.7
131 56.0 30.1 101.0 46.2 1195.0 71.0 37.8 41.4
130 49.3 18.6 100.0 45.2 1222.0 70.0 38.8 91.5
129 54.7 29.0 99.0 45.1 1168.0 69.0 40.3 126.2
128 49.6 42.5 98.0 43.4 672.0 68.0 39.3 270.0
127 52.0 677.0 97.0 41.9 363.0 67.0 40.0 440.0
126 53.6 453.0 96.0 43.8 172.5 66.0 36.5 298.0
125 46.3 1362.0 95.0 45.8 205.0 65.0 37.0 209.0
124 42.7 990.0 94.0 45.7 297.0 64.0 32.8 150.4
123 55.5 914.0 93.0 45.9 450.0 63.0 36.7 88.6
122 55.9 599.0 92.0 44.8 498.0 62.0 35.7 99.5
121 53.0 177.6 91.0 41.0 196.8 61.0 35.2 99.9
120 52.1 73.0 90.0 40.9 112.4 60.0 35.9 83.3
119 54.8 34.4 89.0 43.4 86.1 59.0 35.3 75.1
118 56.0 23.9 88.0 45.2 51.5 58.0 37.9 70.0
117 53.6 14.3 87.0 42.9 35.1 57.0 37.4 76.2
116 54.9 1297.0 86.0 41.5 28.9 56.0 37.4 74.0
115 52.0 685.0 85.0 36.7 79.7 55.0 35.9 50.2
114 45.3 339.0 84.0 43.4 49.0
113 46.4 218.0 83.0 43.2 32.9
112 45.8 170.0 82.0 43.3 30.8

113
BH Name: BH24 Depth of Borehole: 130m Date: 28/9/1997
Measurement Method: Apparent Resistively (No.2M 1.9A) Self potential Logging Vsp (mv)
Depth (m) Vsp (mv) Ps (Ƿm ) Depth (m) Vsp (mv) Ps Depth (m) Vsp (mv) Ps (Ƿm )
(Ƿm )
130 39.1 132.5 104.0 29.0 34.1 79.0 21.5 1853.0
129 28.1 117.2 103.0 28.2 93.3 78.0 19.0 1025.0
128 37.6 137.6 102.0 29.1 76.3 77.0 19.4 351.0
127 36.5 156.5 101.0 26.6 74.9 76.0 21.0 278.0
126 35.2 168.3 100.0 27.3 87.6 75.0 22.6 16.3
125 34.2 173.2 99.0 27.2 88.6 74.0 23.0 154.2
124 31.9 157.2 98.0 27.1 443.0 73.0 22.4 144.5
123 32.4 115.9 97.0 26.7 1717.0 72.0 25.9 107.2
122 31.4 65.4 96.0 24.3 1766.0 71.0 28.3 59.3
121 32.8 33.1 95.0 23.3 1379.0 70.0 23.4 36.2
120 33.5 25.8 94.0 24.5 2100.0 69.0 21.7 15.5
119 32.1 11.1 93.0 25.6 2400.0 68.0 25.0 15.1
118 30.9 12.8 92.0 23.4 2100.0 67.0 22.3 8.4
117 30.7 10.3 91.0 22.1 1289.0 66.0 21.8 16.3
116 31.1 13.1 90.0 22.7 946.0 65.0 22.0 15.9
115 30.8 18.0 89.0 23.4 1475.0 64.0 22.0 16.4
114 30.1 30.7 88.0 21.8 1512.0 63.0 20.9 18.1
113 30.8 117.0 87.0 21.1 1222.0 62.0 21.1 18.9
112 29.9 3160.0 86.0 22.0 1000.0 61.0 22.4 12.3
111 29.2 1988.0 85.0 22.8 968.0 60.0 22.6 36.9
110 27.2 1835.0 84.0 22.9 2260.0 59.0 22.9 250.0
109 28.9 1362.0 83.0 23.2 3500.0 58.0 19.8 944.0
108 28.2 1018.0 82.0 20.2 2940.0 57.0 20.4 599.0
107 28.5 781.0 81.0 20.4 2480.0 56.0 18.3 479.0
106 28.1 15.5 80.0 21.9 2120.0
105 27.3 93.4

114
Appendix 3 Hydrostratigraphic cross-section on the wellfield

North-south well log correlation in Akaki wellfield (2.2 km)

115
Appendix 4 E-W well log correlation of catchment (Dereje Nigusse, 2003)

116
Appendix 5 How to correct ASTER DEM with topographic map
The relationship between the ground elevation and the elevation from ASTER DEM is shown in figure 2.
The elevations from ASTER DEM and those from the topographic map show high correlation with R2 of
0.97. Before applying the elevation extracted from the DEM for further analysis, the elevation is corrected
by the regression equation obtained by the built comparison to the ground control point elevations and the
extracted elevation. The regression equation obtained from the comparison is X = (Y- 161.65)/ (0.8992),
where Y is the elevation from ASTER and X is the elevation from Topomap of the area. The correction
process of the elevation from the ASTER is accomplished by applying this formula in the Map calculation
of ILWIS. After correction, the required array of elevations at the well locations is extracted by applying
the map value function of ILWIS to the ASTER DEM. The following steps illustrate how to correct the
digital elevation model before determining rainfall distribution in the catchment

1. Georeferencing Topo map

X Y Row Col Active DRow DCol
474000 968000 6965 2603 True 0.15 0.21
490000 990000 1823 6453 True -0.17 -0.43
499000 968000 7061 8505 True 0.01 -0.06
490000 980000 4188 6420 True -0.26 -0.16
499000 994000 912 8592 True 0.26 0.44
Sigma = RMSE = 0.402

2. exteraction of elevation from topo map and correcting ASTER DEM with topo map
Name Coordinate DEM_elevation (m) TOPO_elevation (m)
1 ( 491044.00, 973800.00) 2071 2215
2 ( 492407.00, 979125.00) 2232 2290
3 ( 495527.00, 983711.00) 2992 3100
4 ( 493090.00, 984169.00) 2627 2685
5 ( 492809.00, 983498.00) 2426 2675
6 ( 489007.00, 981390.00) 2293 2375
7 ( 484607.00, 984201.00) 2232 2295
8 ( 482944.00, 986524.00) 2239 2285
9 ( 476446.00, 991839.00) 2306 2323
10 ( 476602.00, 994125.00) 2295 2338
11 ( 477319.00, 981117.00) 2038 2080
12 ( 478227.00, 975893.00) 2021 2080
13 ( 477467.00, 970090.00) 2279 2380
14 ( 480681.00, 969888.00) 2329 2475
15 ( 484831.00, 967646.00) 1890 1900
16 ( 491046.00, 973802.00) 2071 2215
17 ( 498704.00, 981218.00) 2305 2320
18 ( 498424.00, 993224.00) 2421 2460
19 ( 481077.00, 988008.00) 2167 2180
20 ( 474577.00, 971828.00) 2074 2120

117
Points used to correct the ASTER DEM

correlation between elevations of ASTER DEM and Topo maps

3200

3000 y = 0.8992x + 161.65

R2 = 0.9701
ASTER_DEM_elevation (m.asl)

2800

2600

2400

2200

2000

1800
1800 2000 2200 2400 2600 2800 3000 3200
Topo elevation(m.asl)
Scatter plot of ground elevation vs elevation from ASTER DEM

3. Corrected DEM for Akaki catchment

Corrected DEM of Akaki catchment

118
Appendix 6 Chloride concentration in Akaki catchment (2007 – 2008)
sample ID X Y Groundwater Cl (mg/l) Average rainfall Cl (mg/l) Year
G-3 462754 995769 0.6 0.95 2007
G-4 462788 992425 1.3 0.95 2007
G-5 503833 1012823 17.6 0.95 2007
G-6 501208 1013728 1.8 0.95 2007
G-7 503448 1012814 3.2 0.95 2007
G-10 481266 981756 8 0.95 2007
G-11 479544 981260 10.8 0.95 2007
G-13 493917 1005045 1.2 0.95 2007
G-14 493167 1004737 3.5 0.95 2007
G-15 491200 1004464 2.9 0.95 2007
G-18 473369 1005062 1.3 0.95 2007
G-19 456121 1002559 12.7 0.95 2007
G-20 458156 998337 1.5 0.95 2007
G-21 464365 988869 6.7 0.95 2007
G-26 479954 977334 6 0.95 2007
G-27 479700 976927 6.4 0.95 2007
G-28 478347 976750 8.2 0.95 2007
G-42 479031 977601 10 0.95 2007
EP-04 479118.5 977807.4 11.7 0.7 2008
BH-05 476680.7 975819.8 4.9 0.7 2008
BH-01 478061.4 975076.1 3.2 0.7 2008
BH-10 479155.1 976241.9 5.0 0.7 2008
F3 479779.2 981769.5 11.7 0.7 2008
Iron & steel BH 476743.0 980936.4 9.2 0.7 2008
Food Factory BH 476760.5 980804.8 23.7 0.7 2008
Spare Parts BH 478347.8 977935.0 6.6 0.7 2008

119
Appendix 7 Plots of time-drawdown curve for selected bore holes

120
Appendix 8 Semi variogram models for aquifer parameter determination

Semivariogram of Guasian model for water level interpolation (nugget = 0, sill = 10, 0000, range = 20000)

Semivariogram of wave model for log10(specific capacity) interpolation (nugget = 0.5, sill = 0.75, range =
5000)

121
Appendix 9 Some of model input parameters to Akaki catchment model
Twenty five drains imposed to the model domain
Layer Row Col Elevation Conductance Drain no.
1 17 101 2597 560 1
1 18 101 2597 560 2
1 18 102 2597 560 3
1 19 101 2597 560 4
1 19 102 2590 560 5
1 20 102 2597 560 6
1 20 103 2597 560 7
1 21 102 2597 560 8
1 21 103 2597 560 9
1 22 102 2597 560 10
1 33 66 2664 178 11
1 33 69 2658 178 12
1 37 38 2678 20 13
1 38 38 2668 20 14
1 39 49 2634 144 15
1 43 18 2649 500 16
1 43 19 2649 500 17
1 43 29 2642 28 18
1 44 17 2649 500 19
1 44 18 2649 500 20
1 44 19 2640 500 21
1 44 20 2649 500 22
1 45 17 2649 500 23
1 96 72 2113 120 24
1 97 69 2103 76 25

Ninteen head dependent boundary nodes imposed to the Akaki catchment model area
Layer Row Col GHB elevation Conductance Bound no.
1 128 31 1938 84 1
1 128 32 1938 84 2
1 129 32 1938 84 3
1 129 33 1938 84 4
1 129 34 1938 84 5
1 129 37 1938 84 6
1 129 38 1938 84 7
1 130 34 1938 84 8
1 130 35 1938 84 9
1 130 36 1938 84 10
1 130 37 1938 84 11
1 130 38 1900 84 12
1 130 39 1938 84 13
1 130 40 1938 84 14
1 131 41 1900 84 15
1 131 45 1900 84 16
1 132 42 1900 84 17
1 132 43 1900 84 18
1 132 44 1900 84 19
One hundred fifty wells applied to the Akaki catchment model

122
Well No. Row Column Stress rates (m3/day) Layer
1 16 105 -168 1
2 16 128 -320 1
3 17 97 -108 1
4 18 108 -235 1
5 19 106 -244 1
6 19 108 -345 1
7 19 126 -175 1
8 21 105 -529 1
9 21 125 -231 1
10 22 104 -349 1
11 22 105 -428 1
12 25 118 -210 1
13 25 120 -145 1
14 30 105 -420 1
15 32 104 -420 1
16 35 121 -321 1
17 36 103 -370 1
18 37 94 -315 1
19 37 98 -450 1
20 40 51 -115 1
21 40 109 -425 1
22 41 32 -34 1
23 41 52 -87 1
24 42 33 -36 1
25 43 26 -86 1
26 43 32 -216 1
27 43 89 -560 1
28 43 90 -1321 1
29 44 35 -12 1
30 45 39 -178 1
31 45 53 -42 1
32 46 34 -30 1
33 46 39 -32 1
34 46 40 -54 1
35 46 43 -62 1
36 46 53 -43 1
37 46 54 -104 1
38 46 56 -316 1
39 46 84 -83 1
40 47 36 -40 1
41 47 66 -3 1
42 48 35 -15 1
43 49 59 -56 1
44 49 74 -126 1
45 50 52 -80 1
46 53 18 -65 1
47 53 60 -1700 1
48 53 80 -346 1
49 53 81 -1642 1
50 54 61 -533 1
51 55 17 -1700 1
52 55 47 -48 1
53 55 73 -200 1
54 57 41 -10 1
55 57 54 0 1
56 57 65 -124 1
57 58 36 -50 1
58 58 40 -43 1
59 58 43 -250 1
60 58 50 -90 1
61 59 43 -115 1
62 59 45 -33 1
63 59 47 -101 1
64 59 51 -45 1
65 59 52 -230 1
66 60 47 -340 1
67 60 49 -97 1
68 60 55 -58 1
69 61 28 -108 1
70 61 46 -210 1
71 61 48 -73 1
72 61 78 -87 1
73 62 40 -41 1
74 62 46 -298 1
75 63 63 -12 1

123
Well No. Row Column Stress rates (m3/day) Layer
76 64 41 -78 1
77 64 44 -55 1
78 64 46 -20 1
79 64 61 -15 1
80 64 83 -1000 1
81 65 33 -162 1
82 65 43 -47 1
83 65 46 -343 1
84 65 47 -342 1
85 65 58 -45 1
86 66 28 -288 1
87 66 30 -1555 1
88 66 34 -80 1
89 66 39 -90 1
90 66 49 -108 1
91 67 37 -43 1
92 67 51 -129 1
93 67 52 -20 1
94 68 36 -76 1
95 69 41 -45 1
96 70 45 -30 1
97 71 44 -34 1
98 72 30 -99 1
99 73 45 -1440 1
100 74 43 -432 1
101 74 52 -189 1
102 74 53 -227 1
103 74 76 -432 1
104 76 35 -321 1
105 77 29 -332 1
106 77 52 -210 1
107 78 29 -214 1
108 78 47 -583 1
109 78 53 -180 1
110 79 29 -320 1
111 79 36 -312 1
112 80 52 -253 1
113 81 53 -209 1
114 81 54 -216 1
115 84 56 -576 1
116 85 52 -150 1
117 86 57 -176 1
118 87 58 -149 1
119 89 57 -645 1
120 92 54 -176 1
121 93 56 -220 1
122 93 70 -150 1
123 94 68 -150 1
124 95 60 -137 1
125 95 62 -150 1
126 95 72 -547 1
127 96 51 -187 1
128 96 68 -1901 1
129 97 57 -123 1
130 97 60 -250 1
131 99 65 -2400 1
132 101 62 -137 1
133 102 63 -210 1
134 104 64 -137 1
135 105 64 -1340 1
136 105 68 -2500 1
137 105 69 -500 1
138 106 64 -5216 1
139 106 67 -7808 1
140 106 68 -5953 1
141 107 61 -6000 1
142 107 64 -6912 1
143 107 65 -7517 1
144 108 61 -6000 1
145 108 63 -6496 1
146 108 67 -3616 1
147 109 62 -6000 1
148 109 63 -7517 1
149 112 65 -7517 1
150 116 52 -390 1

124
Appendix 10 Wells, head dependent boundaries, & rivers applied to wellfield
i) well abstructions imposed to welfield model
3
Well no. Row Col Stress rate (m /day) Layer
1 16 105 -168 1
2 16 128 -320 1
3 17 97 -108 1
4 18 108 -235 1
5 19 106 -244 1
6 19 108 -345 1
7 19 126 -175 1
8 21 105 -529 1
9 21 125 -231 1
10 22 104 -349 1
11 22 105 -428 1
12 25 118 -210 1
13 25 120 -145 1
14 30 105 -420 1
15 32 104 -420 1
16 35 121 -321 1
17 36 103 -370 1
18 37 94 -315 1
19 37 98 -450 1
20 40 51 -115 1
21 40 109 -425 1
22 41 32 -34 1
23 41 52 -87 1
24 42 33 -36 1
25 43 26 -86 1
26 43 32 -216 1
27 43 89 -560 1
28 43 90 -1321 1
29 44 35 -12 1
30 45 39 -178 1
31 45 53 -42 1
32 46 34 -30 1
33 46 39 -32 1
34 46 40 -54 1
35 46 43 -62 1
36 46 53 -43 1
37 46 54 -104 1
38 46 56 -316 1
39 46 84 -83 1
40 47 36 -40 1
41 47 66 -3 1
42 48 35 -15 1

125
ii) One hundred twelve head dependent boundary nodes used to simulate inflow and out flow flax to
and out of model domain of the wellfield
2
Layer Row Column Elevation (m) Conductance (m /day)
1 1 1 2167 10.00
1 1 2 2164 10.00
1 1 3 2160 10.00
1 1 4 2154 10.00
1 1 5 2147 10.00
1 1 6 2140 10.00
1 1 7 2135 10.00
1 1 8 2128 15.00
1 1 9 2121 15.00
1 1 10 2113 15.00
1 1 11 2107 15.00
1 1 12 2110 15.00
1 1 13 2102 15.00
1 1 14 2100 15.00
1 1 15 2199 15.00
1 1 16 2097 15.00
1 1 17 2096 15.00
1 1 18 2095 15.00
1 1 19 2093 15.00
1 1 20 2093 15.00
1 1 21 2093 15.00
1 1 22 2093 15.00
1 1 23 2093 15.00
1 1 24 2094 15.00
1 1 25 2096 15.00
1 1 26 2098 15.00
1 1 27 2100 15.00
1 1 28 2102 15.00
1 1 29 2104 15.00
1 1 30 2107 15.00
1 1 31 2109 15.00
1 1 32 2112 10.00
1 1 33 2115 10.00
1 1 34 2120 10.00
1 1 35 2126 10.00
1 1 36 2132 1.21
1 1 37 2137 1.21
1 1 38 2142 1.21
1 1 39 2146 1.21
1 1 40 2151 1.21
1 1 41 2155 1.21
1 1 42 2157 1.21
1 1 43 2157 1.04
1 2 1 2167 10.00
1 3 1 2161 10.00
1 4 1 2161 10.00
1 5 1 2154 10.00
1 6 1 2154 10.00
1 7 1 2147 10.00
1 8 1 2147 10.00
1 9 1 2140 10.00
1 10 1 2140 10.00
1 11 1 2132 10.00
1 12 1 2132 10.00
1 13 1 2125 10.00
1 14 1 2125 10.00
1 15 1 2117 10.00
1 16 1 2117 10.00

126
 2
Layer Row Column Elevation (m) Conductance (m /day)
1 17 1 2109 10.00
1 18 1 2109 10.00
1 19 1 2100 10.00
1 20 1 2100 10.00
1 21 1 2091 10.00
1 22 1 2091 10.00
1 57 6 1988 42.00
1 57 7 1990 42.50
1 57 8 1990 30.00
1 57 9 1990 30.00
1 57 10 1990 30.00
1 57 15 2019 22.00
1 57 16 2020 30.65
1 57 17 2021 18.20
1 57 18 2021 24.30
1 58 6 1988 42.00
1 58 7 1990 42.50
1 58 8 1990 30.00
1 58 9 1990 30.00
1 58 10 1990 30.00
1 58 15 2019 22.00
1 58 16 2020 30.65
1 58 17 2021 18.20
1 58 18 2021 24.30
1 59 4 2000 43.00
1 59 5 1995 42.50
1 59 11 2005 17.25
1 59 12 2005 17.25
1 59 13 2013 14.90
1 59 14 2020 17.95
1 59 19 2022 16.82
1 59 20 2022 16.82
1 59 21 2022 16.82
1 59 22 2023 16.82
1 60 3 2000 43.00
1 60 4 2000 43.00
1 60 5 1995 42.50
1 60 11 2005 17.25
1 60 12 2005 17.25
1 60 13 2013 14.90
1 60 14 2020 17.95
1 60 19 2022 16.82
1 60 20 2022 16.82
1 60 21 2022 16.82
1 60 22 2023 16.82
1 61 3 2000 43.00
1 61 22 2021 16.82
1 62 2 2000 43.00
1 62 3 2000 43.00
1 62 22 2021 16.82
1 63 22 2023 16.82
1 64 22 2023 16.82
1 65 22 2023 16.82
1 66 22 2023 16.82

127
 iii) one hundred twenty two rivers reaches for Akaki wellfield
2
Layer Row Column River Stage(m) Conductance(m /day) Bottom Elevation (m).
1 1 11 2104.00 132.50 2103.50
1 1 23 2090.00 45.00 2089.50
1 2 11 2103.00 132.50 2102.00
1 2 23 2089.00 45.00 2088.00
1 3 11 2101.00 132.50 2100.00
1 3 23 2088.00 45.00 2087.00
1 3 24 2086.00 75.00 2085.00
1 4 11 2101.00 132.50 2100.00
1 4 23 2088.00 45.00 2087.00
1 4 24 2086.00 75.00 2085.00
1 5 11 2100.00 110.50 2099.00
1 5 24 2084.00 91.00 2083.00
1 5 25 2081.00 68.00 2080.00
1 6 11 2100.00 110.50 2099.00
1 6 24 2084.00 91.00 2083.00
1 6 25 2081.00 68.00 2080.00
1 7 12 2072.00 136.50 2071.00
1 7 24 2077.00 55.50 2076.00
1 7 25 2079.00 72.50 2078.00
1 8 12 2072.00 136.50 2071.00
1 8 24 2077.00 55.50 2076.00
1 8 25 2079.00 72.50 2078.00
1 9 12 2071.00 68.00 2070.00
1 9 23 2076.00 81.50 2075.00
1 10 12 2071.00 68.00 2070.00
1 10 23 2076.00 81.50 2075.00
1 11 12 2070.00 82.50 2069.00
1 11 23 2075.00 144.50 2074.00
1 12 12 2070.00 82.50 2069.00
1 12 23 2075.00 144.50 2074.00
1 13 12 2059.00 67.00 2058.00
1 13 23 2073.00 115.00 2072.00
1 14 12 2059.00 67.00 2058.00
1 14 23 2073.00 115.00 2072.00
1 15 13 2058.00 73.50 2057.00
1 15 22 2072.00 144.00 2071.00
1 16 13 2058.00 73.50 2057.00
1 16 22 2072.00 144.00 2071.00
1 17 12 2056.00 60.00 2055.00
1 17 13 2057.00 71.50 2056.00
1 17 20 2060.00 73.00 2059.00
1 17 21 2062.00 150.00 2061.00
1 17 22 2064.00 92.00 2063.00
1 18 12 2056.00 60.00 2055.00
1 18 13 2057.00 71.50 2056.00
1 18 20 2060.00 73.00 2059.00
1 18 21 2062.00 150.00 2061.00
1 18 22 2064.00 92.00 2063.00
1 19 12 2054.00 117.50 2053.00
1 19 20 2059.00 110.00 2058.00
1 20 12 2054.00 117.50 2053.00
1 20 20 2059.00 110.00 2058.00
1 21 10 2021.00 120.00 2020.00
1 21 11 2022.00 63.00 2021.00
1 21 19 2055.00 175.00 2054.00
1 21 20 2056.00 142.50 2055.00
1 21 21 2057.00 70.00 2056.00
1 22 10 2021.00 120.00 2020.00
1 22 11 2022.00 63.00 2021.00
1 22 19 2055.00 175.00 2054.00
1 22 20 2056.00 142.50 2055.00
1 22 21 2057.00 70.00 2056.00
1 23 10 2019.00 116.00 2018.00

128
 2
Layer Row Column River Stage(m) Conductance(m /day) Bottom Elevation (m).
1 23 18 2054.00 124.00 2053.00
1 24 10 2019.00 116.00 2018.00
1 24 18 2054.00 124.00 2053.00
1 25 9 2017.00 56.00 2016.00
1 25 10 2018.00 52.50 2017.00
1 25 19 2052.00 225.00 2051.00
1 26 9 2017.00 56.00 2016.00
1 26 10 2018.00 52.50 2017.00
1 26 19 2052.00 225.00 2051.00
1 27 9 2015.00 67.00 2014.00
1 27 20 2050.00 190.00 2049.00
1 27 21 2048.00 125.00 2047.00
1 28 9 2015.00 67.00 2014.00
1 28 20 2050.00 190.00 2049.00
1 28 21 2048.00 125.00 2047.00
1 29 9 2014.00 123.00 2013.00
1 29 21 2046.00 183.00 2045.00
1 30 9 2014.00 123.00 2013.00
1 30 21 2046.00 183.00 2045.00
1 31 21 2044.00 151.50 2043.00
1 32 21 2044.00 151.50 2043.00
1 33 20 2040.00 101.50 2039.00
1 33 21 2042.00 169.00 2041.00
1 34 20 2040.00 101.50 2039.00
1 34 21 2042.00 169.00 2041.00
1 35 19 2038.00 165.00 2037.00
1 36 19 2038.00 165.00 2037.00
1 37 19 2036.00 207.00 2035.00
1 38 19 2036.00 207.00 2035.00
1 39 16 2028.00 150.00 2027.00
1 39 17 2030.00 173.50 2029.00
1 39 18 2032.00 160.00 2031.00
1 39 19 2034.00 160.00 2033.00
1 40 16 2028.00 150.00 2027.00
1 40 17 2030.00 173.50 2029.00
1 40 18 2032.00 160.00 2031.00
1 40 19 2034.00 160.00 2033.00
1 41 16 2026.00 241.00 2025.00
1 42 16 2026.00 241.00 2025.00
1 43 11 2010.00 181.00 2009.00
1 43 12 2012.00 197.50 2011.00
1 43 15 2020.00 125.00 2019.00
1 43 16 2024.00 322.00 2023.00
1 44 11 2010.00 181.00 2009.00
1 44 12 2012.00 197.50 2011.00
1 44 15 2020.00 125.00 2019.00
1 44 16 2024.00 322.00 2023.00
1 45 10 2008.00 160.00 2007.00
1 45 13 2014.00 190.00 2013.00
1 45 14 2016.00 160.00 2015.00
1 45 15 2017.00 193.00 2016.00
1 46 10 2008.00 160.00 2007.00
1 46 13 2014.00 190.00 2013.00
1 46 14 2016.00 160.00 2015.00
1 46 15 2017.00 193.00 2016.00
1 47 9 2006.00 263.00 2005.00
1 47 10 2007.00 262.00 2006.00
1 48 9 2006.00 263.00 2005.00
1 48 10 2007.00 262.00 2006.00

129
 Well abstraction data of the wellfield
BH Name Depth to top of 14'' screen Depth SWL(m) Phase Pump diam Pump flow (lit./sec.) well abstractions (m3/day)
BH 6 75 67.82 1 6'' 10 720
BH 7 79 67.42 1 6'' 10 720
BH 8 83 67.20 1 6'' 10 720
BH 9 78 58.70 1 10'' 50 3,600
BH12 76 51.53 1 10'' 50 3,600
BH14 78 59.18 1 10'' 50 3,600
BH16 78 47.98 1 10'' 50 3,600
BH17 77 45.94 1 10'' 50 3,600
BH18 82 54.08 1 10'' 50 3,600
BH21 71 44.68 1 10'' 50 3,600
BH22 58 47.90 1 6'' 10 720
BH1 69 58.50 2 6'' 10 720
BH2 63 52.60 2 6'' 10 720
BH3b 63 63.11 2 6'' 10 720
BH4 65 47.90 2 6'' 10 720
BH5b 75 48.85 2 10'' 50 3,600
BH10 76 72.30 2 6'' 10 720
BH11 66 61.14 2 6'' 10 720
BH13 47 51.17 2 6'' 10 720
BH19 71 51.45 2 10'' 50 3,600
BH 20 73 50.05 2 10'' 50 3,600
BH 23 68 45.02 2 10'' 50 3,600
BH 24 62 42.92 2 10'' 50 3,600
BH 25 80 41.92 2 10'' 50 3,600
BH 26 57 50.95 2 6'' 10 720
Total (m3/day) 55,440

130
Appendix 11 Akaki wellfield monitoring wells
No Date MW-01 MW-02 MW-03 MW-04 P-01 TW-01 time in days
1 8/10/2002 43.9 37.49 43.09 53.69 50.56 51.41 1
2 25/10/2002 43.96 37.56 43.13 53.74 50.59 51.47 17
3 9/11/2002 43.97 37.57 43.16 53.77 50.64 51.47 32
4 5/6/2003 44.82 38.38 43.2 53.82 50.88 51.33 241
5 11/6/2003 44.88 38.43 43.26 53.87 52.43 51.37 248
6 21/06/2003 44.9 38.47 43.3 53.93 51.03 51.4 258
7 28/06/2003 44.94 38.49 43.3 53.93 50.99 51.43 265
8 5/7/2003 44.95 38.52 43.37 54 52.21 51.42 272
9 12/7/2003 45.03 38.61 43.45 54.07 52.25 51.41 279
10 19/07/2003 45.04 38.64 43.47 54.1 52.24 51.48 286
11 26/07/2003 45.08 38.68 43.45 54.15 52.28 51.49 293
12 2/8/2003 45.13 38.73 43.68 54.21 52.31 51.51 300
13 9/8/2003 45.18 38.78 43.66 54.27 52.34 51.55 307
14 16/08/2003 45.23 38.85 43.7 54.33 52.37 51.57 314
15 23/08/2003 45.26 38.88 43.7 54.33 52.3 51.6 321
16 30/08/2003 45.29 38.9 43.7 54.33 51.37 51.6 328
17 6/9/2003 45.31 38.93 43.75 54.38 52.3 51.66 336
18 13/09/2003 45.34 38.97 43.81 54.43 52.3 51.65 343
19 20/09/2003 45.38 39.02 43.88 54.5 51.58 51.7 350
20 18/10/2003 45.54 39.19 44.06 54.69 51.73 51.82 378
21 25/10/2003 45.55 39.21 44.06 54.69 51.76 51.88 385
22 1/11/2003 45.57 39.25 44.1 54.73 51.8 51.88 392
23 8/11/2003 45.62 39.3 44.17 54.78 51.86 51.89 399
24 15/11/2003 45.66 39.34 44.22 54.83 51.88 51.94 406
25 22/11/2003 45.7 39.37 44.23 54.85 51.92 51.98 413
26 29/11/2003 45.73 39.39 44.25 54.86 51.93 52.01 420
27 6/12/2003 45.76 39.42 44.27 54.9 51.95 52.01 427
28 13/12/2003 45.8 39.47 44.3 54.95 51.99 52.01 434
29 20/12/2003 45.84 39.51 44.33 54.99 52.04 52.02 441
30 27/12/2003 45.87 39.56 44.38 55.08 52.13 52.09 448
31 3/1/2004 45.91 39.59 44.46 55.08 52.13 52.13 455
32 10/1/2004 45.93 39.6 44.5 55.1 52.18 52.17 462
33 17/01/2004 45.93 39.6 44.5 55.1 52.18 52.17 469
34 24/01/2004 45.96 39.63 44.55 55.11 52.21 52.16 476
35 31/01/2004 45.98 39.67 44.57 55.13 52.23 52.19 483
36 23/04/2004 54.7 52.6 566
37 24/04/2004 46.48 40.16 45.05 54.68 569
38 16/08/2004 55.27 53.15 683
39 4/10/2004 55.53 53.24 732
40 19/10/2004 55.63 53.32 747
41 14/12/2004 47.54 41.28 55.82 53.96 53.56 803
42 6/3/2005 47.95 41.71 46.66 56.36 53.92 886
43 18/04/2005 41.97 46.85 55.34 54 929
44 12/7/2005 41.76 47.01 53.9 1045
45 26/7/2005 48.69 57.03 54.59 1059
46 2/8/2005 48.67 56.98 54.57 1066
47 11/8/2005 48.71 54.49 1075
48 16/08/2005 48.65 57 54.54 1080
49 23/08/2005 48.71 57.01 54.59 1087
50 30/08/2005 48.61 56.89 54.61 1094
51 6/9/2005 48.71 57.04 54.56 1101
52 15/9/2005 48.59 57.07 54.61 1110

131
Appendix 12 Subcatchments and soil permeability of Akaki area

132
133
Appendix 13 Monthly rainfall data for the water balance model
Intoto Met. Station
Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
1985 14.2 0.0 17.5 96.3 83.7 112.2 270.4 327.7 205.9 58.0 3.3 1.2
1986 0.0 35.7 88.0 197.6 125.4 179.5 180.1 264.2 127.8 36.1 0.0 0.0
1987 0.5 63.4 248.9 82.4 241.3 92.9 196.5 254.4 115.2 21.3 0.8 0.3
1988 9.7 53.4 5.3 144.6 16.6 106.2 277.9 299.3 229.7 59.9 0.0 0.0
1989 4.0 75.8 105.4 133.3 5.8 98.2 409.5 323.4 293.9 11.7 0.0 14.5
1990 0.0 156.6 42.5 117.9 17.6 100.5 325.1 499.7 180.8 32.1 1.5 0.0
1991 12.8 64.9 156.2 37.5 37.0 171.2 258.5 395.2 146.7 1.4 0.0 44.2
1992 52.4 31.5 14.6 42.7 84.6 131.6 247.8 387.0 188.6 42.4 0.0 8.1
1993 15.3 44.6 5.7 147.5 49.0 123.3 266.1 407.6 183.2 28.7 0.0 0.0
1994 0.0 0.0 62.2 56.0 91.0 154.8 336.3 306.9 142.5 0.3 24.4 0.0
1995 0.0 96.3 29.7 186.1 84.0 98.0 291.0 222.1 133.7 5.0 0.2 22.7
1996 31.1 12.2 121.6 78.3 95.2 242.7 387.2 493.9 158.7 1.2 0.6 0.0
1997 21.2 0.0 18.6 77.3 27.4 76.3 256.3 240.8 89.3 88.3 90.0 0.2
1998 25.3 25.3 45.2 47.0 149.5 149.2 369.0 376.3 134.4 44.5 0.0 0.0

Sendafa Met. Station

Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
1985 9.5 0 38.6 158.9 157.9 76.4 394.1 446.1 106 10.8 0 0
1986 0 28.1 117.3 193.7 32.3 164.7 270.9 244.5 143.2 0 0 0
1987 0.2 33.1 128.9 80.5 110.1 55.7 223.6 143.6 105.3 8.7 0 0
1988 0 32.9 0.9 132.3 22.1 104.6 451.1 360.3 198.4 5.2 0 0
1989 18 10.3 43.3 112 21.4 46.3 357.2 339.4 139.9 10.2 0.4 0.6
1990 21.5 190.4 35.7 148.4 38.2 88.5 273.4 470.4 116.3 3.8 2.4 0
1991 15.9 20.5 118.1 0.5 48 94.6 333.4 218.9 134.5 9.4 0 5.6
1992 10.7 46.5 1 29.1 32.5 69.6 257.5 357.9 151.4 55.5 0 0
1993 4.3 105.2 0 78.2 48 94.6 457.2 353 158.4 13.7 0 0
1994 0 0 0 64.1 11 122.7 337.7 184.1 122 0 6.4 0
1995 0 11.4 96.2 116.7 42.4 22.5 230.8 388.8 116.3 0 0 0
1996 69.4 5.6 99.3 78.2 48 180.4 339.2 338.6 111.4 0 0 0
1997 44.5 0 29.4 60 44.8 149.7 303.8 251.1 84.7 72 34.6 0
1998 28.9 23.3 5.8 27 38.2 68.8 359.1 289.7 152 98.9 0 0

Addis Abeba Bole Met. Station

Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sept. Oct. Nov. Dec.
1985 35.1 0 49.1 130.3 92.8 110.9 210.3 260.8 168.6 29.8 0 0.4
1986 0 37.6 56.2 216.6 37.7 175.2 167.9 222.3 107.4 31.6 0 2.5
1987 0 49.1 180.1 85.7 154.6 71.9 155.9 98.1 57 16.6 0 0.4
1988 4.7 33.4 6.7 157.9 34.7 93.2 181.4 265.3 187.3 57.3 0 0
1989 3.4 33.7 58.4 143.3 0 88.1 218.1 318.6 150 36.8 0 7.9
1990 3.2 161.1 60.4 144.5 25.2 48.3 204.2 413.4 143 46.1 2.1 0
1991 0.2 29.6 134.1 15 7.7 105.7 279.4 287.9 123.1 4.4 2.1 0
1992 14.5 28 35 58.6 55 82.2 254.8 223.3 157 64.4 2.2 0.4
1993 11.7 52.1 11.6 168.3 91.5 157.2 209.5 291.7 190.1 24.1 0 0
1994 0 0 52.9 70 29 111.4 242.3 199.3 99.4 0.5 11 0
1995 0 81.3 73.1 133.3 95.9 77.4 165.5 256.9 97 0 0 29.3
1996 19.6 15.8 134.4 96.5 124.6 290.2 346.3 312.7 211.4 0 0.4 0
1997 29.1 0 22.1 66.8 44.8 128 257 160.7 94.7 58.6 3.1 0
1998 63.1 40 43.8 87.8 193.1 111.6 256.8 236.8 185.2 139.5 0 0

Melka Kunture Met. Station

Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
1985 15.8 0 28 36.2 50.2 47.8 270.5 240.2 50.4 19.3 0 0
1986 0 74.6 56 108.8 39.7 136.5 114.5 185.7 84.2 0 0 12.2
1987 0 87.7 132.4 98.3 167.2 74.1 149.5 154 30.5 6.6 0 0
1988 0 26.1 2.7 68.1 10.1 104 189.6 272.3 122.2 8.1 0 0
1989 9.4 44 46.5 73.9 10.8 53.7 272.1 193 66.5 6.5 0 11.7
1990 0 137.2 83.2 125.5 26.6 91.6 292.6 223.7 59.8 0 0 0
1991 0 37.4 131.9 5.3 4.9 132.9 375.1 416.9 129.5 3 0 0
1992 28.4 59.1 14.7 27 74.8 144.3 250.5 223.3 92.8 37 0 4.1
1993 0 72.9 19.8 193.1 70 231.7 491.6 434.6 270.3 50.6 0 0
1994 0 0 125.2 25.7 45.7 70.6 240.2 216.5 132.5 0 8.8 0
1995 11.1 35.4 69.2 74.2 63.5 103.7 250.5 228.3 95.4 19.3 0 0
1996 34.5 0 94.9 113.5 132.9 215 300.8 250.8 91.5 0 0 0
1997 10.5 0 11.6 73.1 24.9 117.8 159.4 139.6 50 66.2 11.1 0
1998 33.3 14.7 109.9 40.9 148.9 103.2 279.1 257.2 133 59.1 0 0

134
Akaki Besseka Met. Station
Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
1985 3.6 0 32.4 72.4 96.6 93.5 302.7 324.1 164.3 1.6 0 0
1986 0 95.4 66.9 148.7 68.2 143.4 189.4 216.5 86.1 9.4 0 0
1987 0 65.6 181.9 80.7 187.7 69.3 202 246.9 81.7 4.4 0 0
1988 0 44.5 0 96 23.8 124.6 255.9 278.1 254.2 35.4 0 0
1989 2.1 63.8 53.8 226.3 7.1 58.6 264.2 301 170.9 37.9 0 0
1990 7.7 120.6 48.4 129.4 37.8 78.9 280.7 222.9 117.3 5.8 1.2 0
1991 0 37.6 62.4 11.6 45.6 90.4 263.7 308.5 113.3 4.4 0 56.5
1992 34.7 24.2 30.5 15.5 25.6 100.4 218.4 276 86.7 43.3 0.2 0
1993 1.2 53.9 60.1 118.4 62.5 116.5 218 251.5 118.3 20.3 0 0
1994 0 0 62.7 72.2 20.2 125 225.1 168.9 106.8 0 8.1 0
1995 0 25.4 63.7 102.1 20.9 95.7 269.2 242.3 79.5 0 0 4.8
1996 15.3 0.3 79.7 38.8 90.5 240.1 292.5 234.1 119 1.9 0 0
1997 27.6 0 29.5 102.7 25.2 57 203.6 203.4 82.5 114.9 10.3 0
1998 32.7 30.2 19.6 69.3 159.9 116.9 207.8 280 118.5 36 0 0

Monthly Potential Evapotranspiration - Addis Ababa Observatory

Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
1985 174.72 171.85 216.95 160.89 178.78 151.43 120.3 126.17 147.51 186.68 171.93 166.98
1986 166.45 154.53 190.67 159.61 184.5 134.58 133.16 133.02 153.57 186.14 182.62 171.96
1987 171.27 161.92 162.89 185.61 169.62 147.67 137.83 158.25 153.82 188.35 173.32 168.35
1988 163.6 180.34 217.7 176.64 201.39 147.79 111.27 131.3 131.29 174.08 164.13 166.25
1989 159.39 158.29 179.67 157.56 196.61 152.74 124.16 142.33 137.63 176.42 168.36 164.54
1990 170.44 136.1 181.81 166.58 184.46 146.28 125.03 127.95 134.43 175.18 158.91 160.25
1991 170.64 159.65 185.52 186.13 195.22 150.02 119.48 128.92 143.8 175.64 158.46 154.09
1992 157.25 159.33 141.77 175.69 184.78 145.61 120.63 110.68 132.46 160.02 149.87 149.97
1993 151.81 139.95 199.45 154.96 165.3 130.27 121.04 126.26 119.15 162.57 157.58 153.01
1994 163.23 165.48 181.47 169.62 184.6 126.07 112.91 114.98 139.19 182.3 153.02 153.09
1995 161.76 158.07 183.33 149.47 180.33 157.18 118.1 123.92 135.68 175.01 160.69 151.85
1996 146.35 169.64 174.83 168.75 169.89 120.48 120.13 125.07 133.91 178.67 153.73 154.58
1997 149.98 167.12 191.26 167.04 196.88 161.22 124.55 133.26 169.84 170.65 147.98 156.03
1998 160.47 157.21 182.89 188.91 171.8 147.82 123.15 127.66 129.16 153.52 160.37 156.45

Derived monthly flow volume (in MCM) of Akaki River at Aba Samuel
Year Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec.
1985 3.7 3.5 3.4 4.2 12.7 7.8 67.9 269.9 88.2 14.7 5.1 6.9
1986 5.4 5.4 9.6 14 7.1 15.5 48.5 119.3 69.5 19.8 8.3 3
1987 3.4 3.7 11.8 18.6 17.6 14 39.2 56.8 15 4.4 3.4 4.7
1988 5.7 6.1 4.9 8.8 5.2 7.4 42.8 136.4 108 20.8 10 5.1
1989 4.4 5.1 4.6 12 7.4 10.6 81.1 253.8 102.4 10.6 6.4 7.3
1990 6.6 14.5 15.2 28.6 8.6 11.5 67.3 292.9 94.1 20.1 7.3 7.1
1991 6.5 7.1 8.7 7 5.2 16.1 96 250.5 189.8 19.5 14.2 15.2
1992 12 13.6 8.3 9.1 9.8 11 62.1 183.6 143.5 18 9 9.7
1993 7.9 10.5 6.7 23.8 17.7 37.9 161 349.5 260.8 27.3 7.7 6.4
1994 5.6 7.1 6.2 6.7 7.7 11.6 66.7 134.4 95.8 12.7 9.4 8.6
1995 8.3 12.6 10.8 20.1 13.4 14.5 56.8 214.4 66.6 9.8 7 7.2
1996 7.4 5.1 10.2 11 16.5 79.5 254 491.3 142 25.4 16.6 15.4
1997 15.5 10.5 11.8 10.5 8.5 14.1 68 133.5 27.5 8.3 5.4 4
1998 5.5 5.1 4.8 5.3 20.8 15.1 140.5 411.6 182.5 43.7 12.2 10.8

135
Appendix 14 Pictures of the area

136
137
138